From 9e25dd708fa7a8b2af354e6901683fd604c4a19a Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Mon, 31 Aug 2026 21:47:13 -0700
Subject: [PATCH 01/16] feat(streaming): carry final response cost on streamed
 usage by default

Streamed responses through the proxy previously exposed no usable cost:
the x-litellm-response-cost header is unreadable mid-stream and the final
usage chunk carried only tokens, priced against an alias model name the
client cannot resolve. The include_cost_in_streaming_usage flag existed
but was off by default and only fixed the wire, not SDK clients.

Stamp usage.cost into the joined streaming response by default wherever a
final usage object is built: the chat-completions stream_chunk_builder,
the native /v1/responses RESPONSE_COMPLETED event, and synthetic response
events. Provider-reported cost always wins over the computed value, and
only positive computed costs are stamped so unpriceable alias responses
keep deferring to the logging object's own calculation. Per-chunk SSE
cost injection (/v1/messages, generateContent, passthrough) stays behind
the flag.

Also normalize non-litellm usage objects in stream_chunk_builder: openai
CompletionUsage lacks Usage.__contains__, so membership probes silently
returned False and client-side rebuilds dropped the wire cost and
recounted token usage locally. Wire token counts and cost now survive.

Resolves LIT-6427
---
 .../streaming_chunk_builder_utils.py          |  8 ++-
 litellm/main.py                               | 22 ++++---
 .../streaming_iterator.py                     | 10 ---
 litellm/responses/streaming_iterator.py       | 46 ++++++-------
 .../test_streaming_chunk_builder_utils.py     | 53 +++++++++++++++
 .../responses/test_streaming_iterator.py      | 52 +++++++++++++++
 tests/test_litellm/test_main.py               | 66 +++++++++++++++++--
 7 files changed, 204 insertions(+), 53 deletions(-)

diff --git a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
index 0e2139d688b..276616f9eee 100644
--- a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
+++ b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
@@ -36,6 +36,8 @@ from litellm.types.utils import (
 from litellm.utils import print_verbose, token_counter
 
 if TYPE_CHECKING:
+    from openai.types.completion_usage import CompletionUsage
+
     from litellm.litellm_core_utils.litellm_logging import Logging
     from litellm.types.litellm_core_utils.streaming_chunk_builder_utils import (
         UsagePerChunk,
@@ -782,7 +784,7 @@ class ChunkProcessor:
 
     @staticmethod
     def _extract_usage_chunk(chunk: "_UsageBearingChunk | ModelResponse | ModelResponseStream") -> Usage | None:
-        usage_chunk: Usage | None = None
+        usage_chunk: Usage | CompletionUsage | None = None
         if hasattr(chunk, "usage") and chunk.usage is not None:
             usage_chunk = chunk.usage
         elif "usage" in chunk:
@@ -794,7 +796,9 @@ class ChunkProcessor:
 
         if isinstance(usage_chunk, dict):
             return Usage(**usage_chunk)
-        return usage_chunk
+        if usage_chunk is None or isinstance(usage_chunk, Usage):
+            return usage_chunk
+        return Usage(**usage_chunk.model_dump())
 
     def _calculate_usage_per_chunk(
         self,
diff --git a/litellm/main.py b/litellm/main.py
index 0c8bff16f81..7ca84226b09 100644
--- a/litellm/main.py
+++ b/litellm/main.py
@@ -8634,6 +8634,16 @@ def _set_stream_builder_response_cost(response: ModelResponse, logging_obj: Opti
     hidden_params["response_cost"] = response_cost
 
 
+def _stamp_streaming_usage_cost(usage: Usage, response: ModelResponse, logging_obj: Optional["Logging"]) -> None:
+    if logging_obj is None:
+        return
+    if isinstance(getattr(usage, "cost", None), (int, float)):
+        return
+    computed_cost: Final = logging_obj._response_cost_calculator(result=response)
+    if isinstance(computed_cost, (int, float)) and computed_cost > 0:
+        setattr(usage, "cost", computed_cost)
+
+
 def stream_chunk_builder(
     chunks: list,
     messages: list | None = None,
@@ -8728,12 +8738,7 @@ def stream_chunk_builder(
                     )
                     break
 
-            if litellm.include_cost_in_streaming_usage and logging_obj is not None:
-                setattr(
-                    usage,
-                    "cost",
-                    logging_obj._response_cost_calculator(result=response),
-                )
+            _stamp_streaming_usage_cost(usage, response, logging_obj)
             _set_stream_builder_response_cost(response, logging_obj)
 
             processor.apply_provider_assembled_streaming_metadata(response, chunks, logging_obj)
@@ -8912,10 +8917,7 @@ def stream_chunk_builder(
                 )
                 break
 
-        # Add cost to usage object if include_cost_in_streaming_usage is True
-        if litellm.include_cost_in_streaming_usage and logging_obj is not None:
-            setattr(usage, "cost", logging_obj._response_cost_calculator(result=response))
-
+        _stamp_streaming_usage_cost(usage, response, logging_obj)
         _set_stream_builder_response_cost(response, logging_obj)
 
         processor.apply_provider_assembled_streaming_metadata(response, chunks, logging_obj)
diff --git a/litellm/responses/litellm_completion_transformation/streaming_iterator.py b/litellm/responses/litellm_completion_transformation/streaming_iterator.py
index 8b1eeb30306..27afff39c0f 100644
--- a/litellm/responses/litellm_completion_transformation/streaming_iterator.py
+++ b/litellm/responses/litellm_completion_transformation/streaming_iterator.py
@@ -1164,16 +1164,6 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
 
     def _emit_response_completed_event(self, litellm_model_response: ModelResponse) -> ResponseCompletedEvent | None:
         if litellm_model_response:
-            # Add cost to usage object if include_cost_in_streaming_usage is True
-            if litellm.include_cost_in_streaming_usage and self.litellm_logging_obj is not None:
-                usage: Final[object] = getattr(litellm_model_response, "usage", None)
-                if usage is not None:
-                    setattr(
-                        usage,
-                        "cost",
-                        self.litellm_logging_obj._response_cost_calculator(result=litellm_model_response),
-                    )
-
             # Transform the response
             responses_api_response: Final = (
                 LiteLLMCompletionResponsesConfig.transform_chat_completion_response_to_responses_api_response(
diff --git a/litellm/responses/streaming_iterator.py b/litellm/responses/streaming_iterator.py
index d070f7758fd..2b4252aa1c5 100644
--- a/litellm/responses/streaming_iterator.py
+++ b/litellm/responses/streaming_iterator.py
@@ -405,23 +405,7 @@ class BaseResponsesAPIStreamingIterator:
                     openai_types.ResponsesAPIStreamEvents.RESPONSE_FAILED,
                 ):
                     self.completed_response = openai_responses_api_chunk
-                    # Add cost to usage object if include_cost_in_streaming_usage is True
-                    if litellm.include_cost_in_streaming_usage and self.logging_obj is not None:
-                        response_obj: Final[ResponsesAPIResponse | None] = getattr(
-                            openai_responses_api_chunk, "response", None
-                        )
-                        if response_obj:
-                            usage_obj: Final[ResponseAPIUsage | None] = getattr(response_obj, "usage", None)
-                            if usage_obj is not None:
-                                try:
-                                    cost: Final[float | None] = self.logging_obj._response_cost_calculator(
-                                        result=response_obj
-                                    )
-                                    if cost is not None:
-                                        setattr(usage_obj, "cost", cost)
-                                except Exception:
-                                    # Best-effort usage cost annotation should not break stream replay.
-                                    pass
+                    _stamp_responses_usage_cost(getattr(openai_responses_api_chunk, "response", None), self.logging_obj)
 
                     if _chunk_type == openai_types.ResponsesAPIStreamEvents.RESPONSE_FAILED:
                         self._handle_logging_failed_response()
@@ -1272,6 +1256,24 @@ def _add_text_like_part_events(
         )
 
 
+def _stamp_responses_usage_cost(
+    response_obj: ResponsesAPIResponse | None, logging_obj: LiteLLMLoggingObj | None
+) -> None:
+    if response_obj is None or logging_obj is None:
+        return
+    usage_obj: Final[ResponseAPIUsage | None] = getattr(response_obj, "usage", None)
+    if usage_obj is None:
+        return
+    if isinstance(getattr(usage_obj, "cost", None), (int, float)):
+        return
+    try:
+        cost: Final[float | None] = logging_obj._response_cost_calculator(result=response_obj)
+    except Exception:
+        return
+    if isinstance(cost, (int, float)) and cost > 0:
+        setattr(usage_obj, "cost", cost)
+
+
 def _build_synthetic_response_events(
     *,
     transformed: ResponsesAPIResponse,
@@ -1279,15 +1281,7 @@ def _build_synthetic_response_events(
     chunk_size: int,
 ) -> list[ResponsesAPIStreamingResponse]:
     openai_types: Final = _get_openai_response_types()
-    if litellm.include_cost_in_streaming_usage and logging_obj is not None:
-        usage_obj: Final = transformed.usage if hasattr(transformed, "usage") else None
-        if usage_obj is not None:
-            try:
-                cost: Final[float | None] = logging_obj._response_cost_calculator(result=transformed)
-                if cost is not None:
-                    setattr(usage_obj, "cost", cost)
-            except Exception:
-                pass
+    _stamp_responses_usage_cost(transformed, logging_obj)
 
     events: Final[list[ResponsesAPIStreamingResponse]] = [
         _build_response_status_event(openai_types.ResponsesAPIStreamEvents.RESPONSE_CREATED, transformed),
diff --git a/tests/test_litellm/litellm_core_utils/test_streaming_chunk_builder_utils.py b/tests/test_litellm/litellm_core_utils/test_streaming_chunk_builder_utils.py
index 8ac050a04f9..bacbcbf132b 100644
--- a/tests/test_litellm/litellm_core_utils/test_streaming_chunk_builder_utils.py
+++ b/tests/test_litellm/litellm_core_utils/test_streaming_chunk_builder_utils.py
@@ -592,6 +592,59 @@ def test_stream_chunk_builder_litellm_usage_chunks():
     assert usage.total_tokens == 77
 
 
+def test_calculate_usage_honors_openai_sdk_completion_usage_chunks():
+    from openai.types.completion_usage import CompletionUsage
+
+    content_chunk = ModelResponseStream(
+        id="chatcmpl-sdk-usage-1",
+        created=1745513206,
+        model="mantle-claude",
+        object="chat.completion.chunk",
+        system_fingerprint=None,
+        choices=[
+            StreamingChoices(
+                finish_reason="stop",
+                index=0,
+                delta=Delta(
+                    provider_specific_fields=None,
+                    content="ok",
+                    role=None,
+                    function_call=None,
+                    tool_calls=None,
+                    audio=None,
+                ),
+                logprobs=None,
+            )
+        ],
+        provider_specific_fields=None,
+        stream_options={"include_usage": True},
+    )
+    usage_chunk = ModelResponseStream(
+        id="chatcmpl-sdk-usage-1",
+        created=1745513207,
+        model="mantle-claude",
+        object="chat.completion.chunk",
+        system_fingerprint=None,
+        choices=[],
+        provider_specific_fields=None,
+        stream_options={"include_usage": True},
+    )
+    usage_chunk.usage = CompletionUsage(
+        prompt_tokens=20, completion_tokens=60, total_tokens=80, cost=0.000704
+    )
+    assert type(usage_chunk.usage) is CompletionUsage
+
+    chunks = [content_chunk, usage_chunk]
+    usage = ChunkProcessor(chunks=chunks).calculate_usage(
+        chunks=chunks, model="mantle-claude", completion_output=""
+    )
+
+    assert usage.prompt_tokens == 20
+    assert usage.completion_tokens == 60
+    assert usage.total_tokens == 80
+    assert getattr(usage, "cost", None) == pytest.approx(0.000704)
+
+
 def test_get_model_from_chunks_azure_model_router():
     """
     Test that _get_model_from_chunks finds the actual model from Azure Model Router chunks.
diff --git a/tests/test_litellm/responses/test_streaming_iterator.py b/tests/test_litellm/responses/test_streaming_iterator.py
index 677faf7f655..9edcaaef034 100644
--- a/tests/test_litellm/responses/test_streaming_iterator.py
+++ b/tests/test_litellm/responses/test_streaming_iterator.py
@@ -326,3 +326,55 @@ def test_run_post_success_hooks_does_not_report_generation_time_as_overhead():
 
     assert iterator.completed_response._hidden_params["_response_ms"] == 10000.0
     assert "litellm_overhead_time_ms" not in iterator.completed_response._hidden_params
+
+
+def _responses_api_response_with_usage() -> ResponsesAPIResponse:
+    from litellm.types.llms.openai import ResponseAPIUsage
+
+    return ResponsesAPIResponse(
+        id="resp_lit6427",
+        created_at=int(datetime(2025, 1, 1).timestamp()),
+        status="completed",
+        model="mantle-claude",
+        object="response",
+        output=[],
+        usage=ResponseAPIUsage(input_tokens=20, output_tokens=60, total_tokens=80),
+    )
+
+
+def test_stamp_responses_usage_cost_stamps_computed_cost():
+    from litellm.responses.streaming_iterator import _stamp_responses_usage_cost
+
+    response = _responses_api_response_with_usage()
+    logging_obj = Mock(spec=LiteLLMLoggingObj)
+    logging_obj._response_cost_calculator.return_value = 0.000704
+
+    _stamp_responses_usage_cost(response, logging_obj)
+
+    assert getattr(response.usage, "cost", None) == pytest.approx(0.000704)
+    logging_obj._response_cost_calculator.assert_called_once_with(result=response)
+
+
+def test_stamp_responses_usage_cost_keeps_provider_reported_cost():
+    from litellm.responses.streaming_iterator import _stamp_responses_usage_cost
+
+    response = _responses_api_response_with_usage()
+    setattr(response.usage, "cost", 0.5)
+    logging_obj = Mock(spec=LiteLLMLoggingObj)
+
+    _stamp_responses_usage_cost(response, logging_obj)
+
+    assert getattr(response.usage, "cost", None) == pytest.approx(0.5)
+    logging_obj._response_cost_calculator.assert_not_called()
+
+
+def test_stamp_responses_usage_cost_survives_calculator_failure():
+    from litellm.responses.streaming_iterator import _stamp_responses_usage_cost
+
+    response = _responses_api_response_with_usage()
+    logging_obj = Mock(spec=LiteLLMLoggingObj)
+    logging_obj._response_cost_calculator.side_effect = RuntimeError("cost map unavailable")
+
+    _stamp_responses_usage_cost(response, logging_obj)
+
+    assert getattr(response.usage, "cost", None) is None
diff --git a/tests/test_litellm/test_main.py b/tests/test_litellm/test_main.py
index 8cf878d05d9..7c2b9d0be05 100644
--- a/tests/test_litellm/test_main.py
+++ b/tests/test_litellm/test_main.py
@@ -3150,8 +3150,8 @@ def _stream_builder_logging_obj() -> LiteLLMLogging:
     return logging_obj
 
 
-def test_stream_chunk_builder_reports_streaming_usage_cost_when_enabled(monkeypatch: pytest.MonkeyPatch):
-    monkeypatch.setattr(litellm, "include_cost_in_streaming_usage", True)
+def test_stream_chunk_builder_stamps_streaming_usage_cost_by_default(monkeypatch: pytest.MonkeyPatch):
+    monkeypatch.setattr(litellm, "include_cost_in_streaming_usage", False)
     chunks: Final = [
         _stream_builder_text_chunk("gpt-4o", "Hello "),
         _stream_builder_text_chunk("gpt-4o", "world.", finish_reason="stop"),
@@ -3168,11 +3168,45 @@ def test_stream_chunk_builder_reports_streaming_usage_cost_when_enabled(monkeypa
     assert response._hidden_params["response_cost"] == pytest.approx(usage_cost)
 
 
-def test_stream_chunk_builder_defers_cost_to_logging_obj_when_usage_cost_absent(monkeypatch: pytest.MonkeyPatch):
-    monkeypatch.setattr(litellm, "include_cost_in_streaming_usage", False)
+def test_stream_chunk_builder_skips_stamp_when_cost_is_unpriceable():
+    import time as time_module
+
+    from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLogging
+
+    logging_obj: Final = LiteLLMLogging(
+        model="us.anthropic.claude-opus-5",
+        messages=[{"role": "user", "content": "hi"}],
+        stream=True,
+        call_type="completion",
+        start_time=time_module.time(),
+        litellm_call_id="stream-builder-alias-unpriceable",
+        function_id="1",
+    )
+    logging_obj.model_call_details["custom_llm_provider"] = "bedrock"
+    logging_obj.optional_params = {}
+    usage_chunk: Final = _stream_builder_text_chunk("bedrock-claude-opus-5", "")
+    usage_chunk.usage = Usage(prompt_tokens=40, completion_tokens=5, total_tokens=45)
+    chunks: Final = [
+        _stream_builder_text_chunk("bedrock-claude-opus-5", "Hello ", finish_reason="stop"),
+        usage_chunk,
+    ]
+
+    response: Final = litellm.stream_chunk_builder(
+        chunks=chunks, messages=[{"role": "user", "content": "hi"}], logging_obj=logging_obj
+    )
+
+    assert response is not None
+    assert getattr(response.usage, "cost", None) is None
+    assert response._hidden_params.get("response_cost") is None
+
+
+def test_stream_chunk_builder_keeps_provider_reported_usage_cost():
+    usage_chunk: Final = _stream_builder_text_chunk("gpt-4o", "")
+    usage_chunk.usage = Usage(prompt_tokens=10, completion_tokens=5, total_tokens=15, cost=0.5)
     chunks: Final = [
         _stream_builder_text_chunk("gpt-4o", "Hello "),
         _stream_builder_text_chunk("gpt-4o", "world.", finish_reason="stop"),
+        usage_chunk,
     ]
 
     response: Final = litellm.stream_chunk_builder(
@@ -3180,4 +3214,26 @@ def test_stream_chunk_builder_defers_cost_to_logging_obj_when_usage_cost_absent(
     )
 
     assert response is not None
-    assert response._hidden_params.get("response_cost") is None
+    assert getattr(response.usage, "cost", None) == pytest.approx(0.5)
+    assert response._hidden_params["response_cost"] == pytest.approx(0.5)
+
+
+def test_stream_chunk_builder_prices_alias_from_openai_sdk_usage_chunk():
+    from openai.types.completion_usage import CompletionUsage
+
+    usage_chunk: Final = _stream_builder_text_chunk("mantle-claude", "")
+    usage_chunk.usage = CompletionUsage(prompt_tokens=20, completion_tokens=60, total_tokens=80, cost=0.000704)
+    assert type(usage_chunk.usage) is CompletionUsage
+    chunks: Final = [
+        _stream_builder_text_chunk("mantle-claude", "Hello "),
+        _stream_builder_text_chunk("mantle-claude", "world.", finish_reason="stop"),
+        usage_chunk,
+    ]
+
+    response: Final = litellm.stream_chunk_builder(chunks=chunks, messages=[{"role": "user", "content": "hi"}])
+
+    assert response is not None
+    assert response.usage.prompt_tokens == 20
+    assert response.usage.completion_tokens == 60
+    assert getattr(response.usage, "cost", None) == pytest.approx(0.000704)
+    assert response._hidden_params["response_cost"] == pytest.approx(0.000704)

From 7ca035f310f891eea216f3162191739f3720bcb1 Mon Sep 17 00:00:00 2001
From: Kris Xia <xiajiayi0506@gmail.com>
Date: Tue, 1 Sep 2026 11:39:33 +0800
Subject: [PATCH 02/16] fix(gemini): return enabled thinking content by default

---
 .../gemini/vertex_and_google_ai_studio_gemini.py     |  2 +-
 .../test_vertex_and_google_ai_studio_gemini.py       | 12 ++++++++++++
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py
index d8b1e7ba17c..69fe5678de9 100644
--- a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py
+++ b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py
@@ -949,7 +949,7 @@ class VertexGeminiConfig(VertexAIBaseConfig, BaseConfig):
         # For Gemini 3+ models, use thinkingLevel instead of thinkingBudget
         if model and VertexGeminiConfig._is_gemini_3_or_newer(model):
             if thinking_enabled:
-                if thinking_budget is None or thinking_budget == 0:
+                if thinking_budget == 0:
                     params["includeThoughts"] = False
                 else:
                     params["includeThoughts"] = True
diff --git a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py
index bd07bec900f..d2788408e09 100644
--- a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py
+++ b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py
@@ -1185,6 +1185,18 @@ def test_vertex_ai_map_thinking_param_with_budget_tokens_0():
     }
 
 
+def test_vertex_ai_map_thinking_param_without_budget_tokens_for_gemini_3():
+    v = VertexGeminiConfig()
+    result = v.map_openai_params(
+        non_default_params={"thinking": {"type": "enabled"}},
+        optional_params={},
+        model="gemini-3.5-flash",
+        drop_params=False,
+    )
+
+    assert result["thinkingConfig"] == {"includeThoughts": True}
+
+
 def test_vertex_ai_map_tools():
     v = VertexGeminiConfig()
     optional_params = {}

From 24ee419c85f6758369e6582250a50490ce1d6819 Mon Sep 17 00:00:00 2001
From: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Date: Tue, 1 Sep 2026 19:28:04 +0000
Subject: [PATCH 03/16] fix(models): registry audit 2026-09-01 for openai
 realtime, mistral aliases, voyage, xai, fireworks

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 ...odel_prices_and_context_window_backup.json | 133 ++++++++++++++----
 model_prices_and_context_window.json          | 133 ++++++++++++++----
 2 files changed, 212 insertions(+), 54 deletions(-)

diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index 27ff525c15e..08add22c998 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -30606,17 +30606,18 @@
     },
     "gpt-realtime-2": {
         "cache_creation_input_audio_token_cost": 4e-07,
+        "cache_read_input_audio_token_cost": 4e-07,
         "cache_read_input_token_cost": 4e-07,
         "input_cost_per_audio_token": 3.2e-05,
         "input_cost_per_image": 5e-06,
         "input_cost_per_token": 4e-06,
         "litellm_provider": "openai",
-        "max_input_tokens": 32000,
-        "max_output_tokens": 4096,
-        "max_tokens": 4096,
+        "max_input_tokens": 128000,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
         "mode": "realtime",
         "output_cost_per_audio_token": 6.4e-05,
-        "output_cost_per_token": 1.6e-05,
+        "output_cost_per_token": 2.4e-05,
         "supported_endpoints": [
             "/v1/realtime"
         ],
@@ -30680,8 +30681,8 @@
         "input_cost_per_token": 6e-07,
         "litellm_provider": "openai",
         "max_input_tokens": 128000,
-        "max_output_tokens": 4096,
-        "max_tokens": 4096,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
         "mode": "realtime",
         "output_cost_per_audio_token": 2e-05,
         "output_cost_per_token": 2.4e-06,
@@ -30713,7 +30714,7 @@
         "input_cost_per_audio_token": 1e-05,
         "input_cost_per_token": 6e-07,
         "litellm_provider": "openai",
-        "max_input_tokens": 128000,
+        "max_input_tokens": 32000,
         "max_output_tokens": 4096,
         "max_tokens": 4096,
         "mode": "realtime",
@@ -33477,19 +33478,21 @@
         "source": "https://mistral.ai/pricing#api-pricing"
     },
     "mistral/magistral-medium-latest": {
-        "input_cost_per_token": 2e-06,
+        "cache_read_input_token_cost": 1.5e-07,
+        "input_cost_per_token": 1.5e-06,
         "litellm_provider": "mistral",
-        "max_input_tokens": 40000,
-        "max_output_tokens": 40000,
-        "max_tokens": 40000,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
         "mode": "chat",
-        "output_cost_per_token": 5e-06,
-        "source": "https://mistral.ai/news/magistral",
+        "output_cost_per_token": 7.5e-06,
+        "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04",
         "supports_assistant_prefill": true,
         "supports_function_calling": true,
         "supports_reasoning": true,
         "supports_response_schema": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "supports_vision": true
     },
     "mistral/magistral-small-2506": {
         "deprecation_date": "2025-11-30",
@@ -33508,19 +33511,21 @@
         "supports_tool_choice": true
     },
     "mistral/magistral-small-latest": {
-        "input_cost_per_token": 5e-07,
+        "cache_read_input_token_cost": 1.5e-08,
+        "input_cost_per_token": 1.5e-07,
         "litellm_provider": "mistral",
-        "max_input_tokens": 40000,
-        "max_output_tokens": 40000,
-        "max_tokens": 40000,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
         "mode": "chat",
-        "output_cost_per_token": 1.5e-06,
-        "source": "https://mistral.ai/pricing#api-pricing",
+        "output_cost_per_token": 6e-07,
+        "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03",
         "supports_assistant_prefill": true,
         "supports_function_calling": true,
         "supports_reasoning": true,
         "supports_response_schema": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "supports_vision": true
     },
     "mistral/magistral-small-1-2-2509": {
         "deprecation_date": "2026-07-31",
@@ -33652,16 +33657,21 @@
         "supports_vision": true
     },
     "mistral/mistral-medium": {
-        "input_cost_per_token": 2.7e-06,
+        "cache_read_input_token_cost": 1.5e-07,
+        "input_cost_per_token": 1.5e-06,
         "litellm_provider": "mistral",
-        "max_input_tokens": 32000,
-        "max_output_tokens": 8191,
-        "max_tokens": 8191,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
         "mode": "chat",
-        "output_cost_per_token": 8.1e-06,
+        "output_cost_per_token": 7.5e-06,
+        "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04",
         "supports_assistant_prefill": true,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
         "supports_response_schema": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "supports_vision": true
     },
     "mistral/mistral-medium-2312": {
         "deprecation_date": "2025-06-16",
@@ -45539,6 +45549,26 @@
         "mode": "rerank",
         "output_cost_per_token": 0.0
     },
+    "voyage/rerank-3": {
+        "input_cost_per_token": 5e-08,
+        "litellm_provider": "voyage",
+        "max_input_tokens": 32000,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
+        "mode": "rerank",
+        "output_cost_per_token": 0.0,
+        "source": "https://docs.voyageai.com/docs/pricing"
+    },
+    "voyage/rerank-3-lite": {
+        "input_cost_per_token": 2e-08,
+        "litellm_provider": "voyage",
+        "max_input_tokens": 32000,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
+        "mode": "rerank",
+        "output_cost_per_token": 0.0,
+        "source": "https://docs.voyageai.com/docs/pricing"
+    },
     "voyage/voyage-2": {
         "input_cost_per_token": 1e-07,
         "litellm_provider": "voyage",
@@ -46790,6 +46820,27 @@
         "supports_vision": true,
         "supports_web_search": true
     },
+    "xai/grok-build-latest": {
+        "cache_read_input_token_cost": 3e-07,
+        "cache_read_input_token_cost_above_200k_tokens": 6e-07,
+        "input_cost_per_token": 2e-06,
+        "input_cost_per_token_above_200k_tokens": 4e-06,
+        "litellm_provider": "xai",
+        "max_input_tokens": 500000,
+        "max_output_tokens": 500000,
+        "max_tokens": 500000,
+        "mode": "chat",
+        "output_cost_per_token": 6e-06,
+        "output_cost_per_token_above_200k_tokens": 1.2e-05,
+        "source": "https://docs.x.ai/developers/models",
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_web_search": true
+    },
     "xai/grok-4.6": {
         "cache_read_input_token_cost": 5e-07,
         "cache_read_input_token_cost_above_200k_tokens": 1e-06,
@@ -57110,6 +57161,34 @@
         "supports_tool_choice": true,
         "supports_vision": false
     },
+    "fireworks_ai/accounts/fireworks/models/glm-5p3-flash": {
+        "cache_read_input_token_cost": 3e-08,
+        "input_cost_per_token": 1.5e-07,
+        "litellm_provider": "fireworks_ai",
+        "max_input_tokens": 1048576,
+        "max_tokens": 1048576,
+        "mode": "chat",
+        "output_cost_per_token": 5e-07,
+        "source": "https://docs.fireworks.ai/serverless/pricing",
+        "supports_function_calling": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
+    "fireworks_ai/accounts/fireworks/models/inkling": {
+        "cache_read_input_token_cost": 1.7e-07,
+        "input_cost_per_token": 1e-06,
+        "litellm_provider": "fireworks_ai",
+        "max_input_tokens": 1048576,
+        "max_tokens": 1048576,
+        "mode": "chat",
+        "output_cost_per_token": 4.05e-06,
+        "source": "https://fireworks.ai/models/fireworks/inkling",
+        "supports_function_calling": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
     "fireworks_ai/accounts/fireworks/models/qwen3-embedding-8b": {
         "input_cost_per_token": 1e-07,
         "output_cost_per_token": 0.0,
diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json
index 27ff525c15e..08add22c998 100644
--- a/model_prices_and_context_window.json
+++ b/model_prices_and_context_window.json
@@ -30606,17 +30606,18 @@
     },
     "gpt-realtime-2": {
         "cache_creation_input_audio_token_cost": 4e-07,
+        "cache_read_input_audio_token_cost": 4e-07,
         "cache_read_input_token_cost": 4e-07,
         "input_cost_per_audio_token": 3.2e-05,
         "input_cost_per_image": 5e-06,
         "input_cost_per_token": 4e-06,
         "litellm_provider": "openai",
-        "max_input_tokens": 32000,
-        "max_output_tokens": 4096,
-        "max_tokens": 4096,
+        "max_input_tokens": 128000,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
         "mode": "realtime",
         "output_cost_per_audio_token": 6.4e-05,
-        "output_cost_per_token": 1.6e-05,
+        "output_cost_per_token": 2.4e-05,
         "supported_endpoints": [
             "/v1/realtime"
         ],
@@ -30680,8 +30681,8 @@
         "input_cost_per_token": 6e-07,
         "litellm_provider": "openai",
         "max_input_tokens": 128000,
-        "max_output_tokens": 4096,
-        "max_tokens": 4096,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
         "mode": "realtime",
         "output_cost_per_audio_token": 2e-05,
         "output_cost_per_token": 2.4e-06,
@@ -30713,7 +30714,7 @@
         "input_cost_per_audio_token": 1e-05,
         "input_cost_per_token": 6e-07,
         "litellm_provider": "openai",
-        "max_input_tokens": 128000,
+        "max_input_tokens": 32000,
         "max_output_tokens": 4096,
         "max_tokens": 4096,
         "mode": "realtime",
@@ -33477,19 +33478,21 @@
         "source": "https://mistral.ai/pricing#api-pricing"
     },
     "mistral/magistral-medium-latest": {
-        "input_cost_per_token": 2e-06,
+        "cache_read_input_token_cost": 1.5e-07,
+        "input_cost_per_token": 1.5e-06,
         "litellm_provider": "mistral",
-        "max_input_tokens": 40000,
-        "max_output_tokens": 40000,
-        "max_tokens": 40000,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
         "mode": "chat",
-        "output_cost_per_token": 5e-06,
-        "source": "https://mistral.ai/news/magistral",
+        "output_cost_per_token": 7.5e-06,
+        "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04",
         "supports_assistant_prefill": true,
         "supports_function_calling": true,
         "supports_reasoning": true,
         "supports_response_schema": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "supports_vision": true
     },
     "mistral/magistral-small-2506": {
         "deprecation_date": "2025-11-30",
@@ -33508,19 +33511,21 @@
         "supports_tool_choice": true
     },
     "mistral/magistral-small-latest": {
-        "input_cost_per_token": 5e-07,
+        "cache_read_input_token_cost": 1.5e-08,
+        "input_cost_per_token": 1.5e-07,
         "litellm_provider": "mistral",
-        "max_input_tokens": 40000,
-        "max_output_tokens": 40000,
-        "max_tokens": 40000,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
         "mode": "chat",
-        "output_cost_per_token": 1.5e-06,
-        "source": "https://mistral.ai/pricing#api-pricing",
+        "output_cost_per_token": 6e-07,
+        "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03",
         "supports_assistant_prefill": true,
         "supports_function_calling": true,
         "supports_reasoning": true,
         "supports_response_schema": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "supports_vision": true
     },
     "mistral/magistral-small-1-2-2509": {
         "deprecation_date": "2026-07-31",
@@ -33652,16 +33657,21 @@
         "supports_vision": true
     },
     "mistral/mistral-medium": {
-        "input_cost_per_token": 2.7e-06,
+        "cache_read_input_token_cost": 1.5e-07,
+        "input_cost_per_token": 1.5e-06,
         "litellm_provider": "mistral",
-        "max_input_tokens": 32000,
-        "max_output_tokens": 8191,
-        "max_tokens": 8191,
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
         "mode": "chat",
-        "output_cost_per_token": 8.1e-06,
+        "output_cost_per_token": 7.5e-06,
+        "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04",
         "supports_assistant_prefill": true,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
         "supports_response_schema": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "supports_vision": true
     },
     "mistral/mistral-medium-2312": {
         "deprecation_date": "2025-06-16",
@@ -45539,6 +45549,26 @@
         "mode": "rerank",
         "output_cost_per_token": 0.0
     },
+    "voyage/rerank-3": {
+        "input_cost_per_token": 5e-08,
+        "litellm_provider": "voyage",
+        "max_input_tokens": 32000,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
+        "mode": "rerank",
+        "output_cost_per_token": 0.0,
+        "source": "https://docs.voyageai.com/docs/pricing"
+    },
+    "voyage/rerank-3-lite": {
+        "input_cost_per_token": 2e-08,
+        "litellm_provider": "voyage",
+        "max_input_tokens": 32000,
+        "max_output_tokens": 32000,
+        "max_tokens": 32000,
+        "mode": "rerank",
+        "output_cost_per_token": 0.0,
+        "source": "https://docs.voyageai.com/docs/pricing"
+    },
     "voyage/voyage-2": {
         "input_cost_per_token": 1e-07,
         "litellm_provider": "voyage",
@@ -46790,6 +46820,27 @@
         "supports_vision": true,
         "supports_web_search": true
     },
+    "xai/grok-build-latest": {
+        "cache_read_input_token_cost": 3e-07,
+        "cache_read_input_token_cost_above_200k_tokens": 6e-07,
+        "input_cost_per_token": 2e-06,
+        "input_cost_per_token_above_200k_tokens": 4e-06,
+        "litellm_provider": "xai",
+        "max_input_tokens": 500000,
+        "max_output_tokens": 500000,
+        "max_tokens": 500000,
+        "mode": "chat",
+        "output_cost_per_token": 6e-06,
+        "output_cost_per_token_above_200k_tokens": 1.2e-05,
+        "source": "https://docs.x.ai/developers/models",
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_web_search": true
+    },
     "xai/grok-4.6": {
         "cache_read_input_token_cost": 5e-07,
         "cache_read_input_token_cost_above_200k_tokens": 1e-06,
@@ -57110,6 +57161,34 @@
         "supports_tool_choice": true,
         "supports_vision": false
     },
+    "fireworks_ai/accounts/fireworks/models/glm-5p3-flash": {
+        "cache_read_input_token_cost": 3e-08,
+        "input_cost_per_token": 1.5e-07,
+        "litellm_provider": "fireworks_ai",
+        "max_input_tokens": 1048576,
+        "max_tokens": 1048576,
+        "mode": "chat",
+        "output_cost_per_token": 5e-07,
+        "source": "https://docs.fireworks.ai/serverless/pricing",
+        "supports_function_calling": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
+    "fireworks_ai/accounts/fireworks/models/inkling": {
+        "cache_read_input_token_cost": 1.7e-07,
+        "input_cost_per_token": 1e-06,
+        "litellm_provider": "fireworks_ai",
+        "max_input_tokens": 1048576,
+        "max_tokens": 1048576,
+        "mode": "chat",
+        "output_cost_per_token": 4.05e-06,
+        "source": "https://fireworks.ai/models/fireworks/inkling",
+        "supports_function_calling": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
     "fireworks_ai/accounts/fireworks/models/qwen3-embedding-8b": {
         "input_cost_per_token": 1e-07,
         "output_cost_per_token": 0.0,

From d3dab8e294b06badb2d34f31ce6aade9380a49ea Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Tue, 1 Sep 2026 12:46:04 -0700
Subject: [PATCH 04/16] fix(rerank): map provider errors with the resolved
 provider on sync and async paths

---
 litellm/rerank_api/main.py                 | 22 ++++++--
 tests/test_litellm/rerank_api/test_main.py | 61 ++++++++++++++++++++++
 2 files changed, 80 insertions(+), 3 deletions(-)

diff --git a/litellm/rerank_api/main.py b/litellm/rerank_api/main.py
index c8f7842aebf..597d1cfb863 100644
--- a/litellm/rerank_api/main.py
+++ b/litellm/rerank_api/main.py
@@ -43,10 +43,17 @@ async def arerank(
     """
     Async: Reranks a list of documents based on their relevance to the query
     """
+    _custom_llm_provider: str | None = None  # rebind-ok: set by the get_llm_provider unpack; read in the except
     try:
         loop: Final = asyncio.get_event_loop()
         kwargs["arerank"] = True
 
+        _, _custom_llm_provider, _, _ = litellm.get_llm_provider(  # rebind-ok: see pre-declaration above
+            model=model,
+            custom_llm_provider=custom_llm_provider,
+            api_base=kwargs.get("api_base", None),
+        )
+
         func: Final = partial(
             rerank,
             model,
@@ -70,7 +77,11 @@ async def arerank(
             response = init_response
         return response
     except Exception as e:
-        raise e
+        raise exception_type(
+            model=model,
+            custom_llm_provider=_custom_llm_provider or custom_llm_provider,
+            original_exception=e,
+        )
 
 
 @client
@@ -115,6 +126,7 @@ def rerank(
     model_info: Final = kwargs.get("model_info", None)
     user: Final = kwargs.get("user", None)
     client: Final = kwargs.get("client", None)
+    _custom_llm_provider: str | None = None  # rebind-ok: set by the get_llm_provider unpack; read in the except
     try:
         _is_async: Final = kwargs.pop("arerank", False) is True
         optional_params: Final = GenericLiteLLMParams(**kwargs)
@@ -127,7 +139,7 @@ def rerank(
 
         (
             model,
-            _custom_llm_provider,
+            _custom_llm_provider,  # rebind-ok: see pre-declaration above
             dynamic_api_key,
             dynamic_api_base,
         ) = litellm.get_llm_provider(
@@ -538,4 +550,8 @@ def rerank(
         return response
     except Exception as e:
         verbose_logger.error("Error in rerank: %s", e)
-        raise exception_type(model=model, custom_llm_provider=custom_llm_provider, original_exception=e)
+        raise exception_type(
+            model=model,
+            custom_llm_provider=_custom_llm_provider or custom_llm_provider,
+            original_exception=e,
+        )
diff --git a/tests/test_litellm/rerank_api/test_main.py b/tests/test_litellm/rerank_api/test_main.py
index 587be59c550..62149c742d6 100644
--- a/tests/test_litellm/rerank_api/test_main.py
+++ b/tests/test_litellm/rerank_api/test_main.py
@@ -111,6 +111,67 @@ def test_together_rerank_honors_api_base(respx_mock: respx.MockRouter):
     assert mock_route.calls[0].request.headers["authorization"] == "Bearer fake-together-key"
 
 
+DASHSCOPE_404_BODY = {
+    "error": {
+        "message": "The model `does-not-exist` does not exist or you do not have access to it.",
+        "type": "invalid_request_error",
+        "param": None,
+        "code": "model_not_found",
+    },
+    "request_id": "mock-request-id",
+}
+
+
+def test_rerank_error_names_provider_and_keeps_body(respx_mock: respx.MockRouter, monkeypatch):
+    """Regression for the rerank error path mapping with the unresolved provider param:
+    a provider 404 surfaced as 'None - ' instead of naming the provider and its error body."""
+    monkeypatch.delenv("DASHSCOPE_API_BASE", raising=False)
+    monkeypatch.delenv("DASHSCOPE_API_BASE_RERANK", raising=False)
+
+    mock_route = respx_mock.post("https://dashscope.example/v1/reranks")
+    mock_route.return_value = httpx.Response(404, json=DASHSCOPE_404_BODY)
+
+    with pytest.raises(litellm.NotFoundError) as exc_info:
+        litellm.rerank(
+            model="dashscope/does-not-exist",
+            query=MARKER_QUERY,
+            documents=[MARKER_DOC],
+            api_key="fake-dashscope-key",
+            api_base="https://dashscope.example/v1",
+        )
+
+    assert mock_route.called
+    assert "DashscopeException" in str(exc_info.value)
+    assert "does not exist or you do not have access to it" in str(exc_info.value)
+    assert "None - " not in str(exc_info.value)
+
+
+@pytest.mark.asyncio
+async def test_arerank_error_is_mapped_to_litellm_exception(respx_mock: respx.MockRouter, monkeypatch):
+    """Regression for arerank's bare re-raise: provider errors escaped as raw
+    provider exception classes instead of the mapped litellm exception contract."""
+    monkeypatch.delenv("DASHSCOPE_API_BASE", raising=False)
+    monkeypatch.delenv("DASHSCOPE_API_BASE_RERANK", raising=False)
+    monkeypatch.setenv("DISABLE_AIOHTTP_TRANSPORT", "True")
+
+    mock_route = respx_mock.post("https://dashscope.example/v1/reranks")
+    mock_route.return_value = httpx.Response(404, json=DASHSCOPE_404_BODY)
+
+    with pytest.raises(litellm.NotFoundError) as exc_info:
+        await litellm.arerank(
+            model="dashscope/does-not-exist",
+            query=MARKER_QUERY,
+            documents=[MARKER_DOC],
+            api_key="fake-dashscope-key",
+            api_base="https://dashscope.example/v1",
+        )
+
+    assert mock_route.called
+    assert "DashscopeException" in str(exc_info.value)
+    assert "does not exist or you do not have access to it" in str(exc_info.value)
+    assert "None - " not in str(exc_info.value)
+
+
 @pytest.mark.asyncio
 async def test_together_rerank_async_honors_env_api_base(respx_mock: respx.MockRouter, monkeypatch):
     """Regression: TOGETHER_AI_API_BASE was honored by chat but ignored by rerank."""

From 386946353ac425c6cbcac28368846dc2ab413bc2 Mon Sep 17 00:00:00 2001
From: Yujong Lee <yujong@berri.ai>
Date: Tue, 1 Sep 2026 14:39:15 -0700
Subject: [PATCH 05/16] fix(vertex): avoid duplicate DeepSeek OCR model
 namespace

---
 .../vertex_ai/ocr/deepseek_transformation.py  |  3 ++-
 tests/ocr_tests/test_ocr_vertex_ai.py         | 20 ++++++++++++++++++-
 2 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/litellm/llms/vertex_ai/ocr/deepseek_transformation.py b/litellm/llms/vertex_ai/ocr/deepseek_transformation.py
index 2603552152d..b57a87c3325 100644
--- a/litellm/llms/vertex_ai/ocr/deepseek_transformation.py
+++ b/litellm/llms/vertex_ai/ocr/deepseek_transformation.py
@@ -177,8 +177,9 @@ class VertexAIDeepSeekOCRConfig(BaseOCRConfig):
             content_item = {"type": "image_url", "image_url": document_url}
 
         # Build DeepSeek OCR request
+        provider_model: Final = model if model.startswith("deepseek-ai/") else f"deepseek-ai/{model}"
         data: Final = {
-            "model": "deepseek-ai/" + model,
+            "model": provider_model,
             "messages": [{"role": "user", "content": [content_item]}],
         }
 
diff --git a/tests/ocr_tests/test_ocr_vertex_ai.py b/tests/ocr_tests/test_ocr_vertex_ai.py
index 1ba5b9d0883..1842eb063a5 100644
--- a/tests/ocr_tests/test_ocr_vertex_ai.py
+++ b/tests/ocr_tests/test_ocr_vertex_ai.py
@@ -5,9 +5,11 @@ Note: Vertex AI OCR automatically converts URLs to base64 data URIs since
 the Vertex AI endpoint doesn't have internet access.
 """
 
-import os
 import json
+import os
 import tempfile
+from typing import Final
+
 import pytest
 from base_ocr_unit_tests import BaseOCRTest
 
@@ -139,3 +141,19 @@ def test_vertex_ai_ocr_routing():
     assert isinstance(
         deepseek_variant, VertexAIDeepSeekOCRConfig
     ), "DeepSeek variant should route to VertexAIDeepSeekOCRConfig"
+
+
+@pytest.mark.parametrize("model", ("deepseek-ocr-maas", "deepseek-ai/deepseek-ocr-maas"))
+def test_deepseek_request_uses_single_provider_namespace(model: str) -> None:
+    from litellm.llms.vertex_ai.ocr.deepseek_transformation import (
+        VertexAIDeepSeekOCRConfig,
+    )
+
+    request: Final = VertexAIDeepSeekOCRConfig().transform_ocr_request(
+        model=model,
+        document={"type": "image_url", "image_url": "data:image/png;base64,AA=="},
+        optional_params={},
+        headers={},
+    )
+
+    assert request.data["model"] == "deepseek-ai/deepseek-ocr-maas"

From d59fcda8af69f5545a8e7c29b29d12362e2bffad Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Tue, 1 Sep 2026 14:47:59 -0700
Subject: [PATCH 06/16] fix(rerank): adopt declared authenticating providers in
 arerank instead of resolving them

get_llm_provider runs the OAuth device flow for github_copilot and chatgpt,
so calling it on the event loop before the executor dispatch let an
authenticated caller block the loop for the length of the polling window.
Adopt the declared provider via declared_authenticating_provider, matching
the metadata callers in utils.py, and only resolve for everything else.
---
 litellm/rerank_api/main.py                 | 19 +++++++++----
 tests/test_litellm/rerank_api/test_main.py | 32 ++++++++++++++++++++++
 2 files changed, 45 insertions(+), 6 deletions(-)

diff --git a/litellm/rerank_api/main.py b/litellm/rerank_api/main.py
index 597d1cfb863..37ca989b8d3 100644
--- a/litellm/rerank_api/main.py
+++ b/litellm/rerank_api/main.py
@@ -6,6 +6,7 @@ from typing import Any, Final, Literal
 
 import litellm
 from litellm._logging import verbose_logger
+from litellm.litellm_core_utils.get_llm_provider_logic import declared_authenticating_provider
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
 from litellm.llms.base_llm.rerank.transformation import BaseRerankConfig
 from litellm.llms.bedrock.rerank.handler import BedrockRerankHandler
@@ -43,16 +44,22 @@ async def arerank(
     """
     Async: Reranks a list of documents based on their relevance to the query
     """
-    _custom_llm_provider: str | None = None  # rebind-ok: set by the get_llm_provider unpack; read in the except
+    _custom_llm_provider: str | None = (
+        None  # rebind-ok: set by the declared-provider guard or the get_llm_provider unpack; read in the except
+    )
     try:
         loop: Final = asyncio.get_event_loop()
         kwargs["arerank"] = True
 
-        _, _custom_llm_provider, _, _ = litellm.get_llm_provider(  # rebind-ok: see pre-declaration above
-            model=model,
-            custom_llm_provider=custom_llm_provider,
-            api_base=kwargs.get("api_base", None),
-        )
+        declared_provider: Final = declared_authenticating_provider(model, custom_llm_provider)
+        if declared_provider is not None:
+            _custom_llm_provider = declared_provider  # rebind-ok: see pre-declaration above
+        else:
+            _, _custom_llm_provider, _, _ = litellm.get_llm_provider(  # rebind-ok: see pre-declaration above
+                model=model,
+                custom_llm_provider=custom_llm_provider,
+                api_base=kwargs.get("api_base", None),
+            )
 
         func: Final = partial(
             rerank,
diff --git a/tests/test_litellm/rerank_api/test_main.py b/tests/test_litellm/rerank_api/test_main.py
index 62149c742d6..2b6cfeda2c2 100644
--- a/tests/test_litellm/rerank_api/test_main.py
+++ b/tests/test_litellm/rerank_api/test_main.py
@@ -172,6 +172,38 @@ async def test_arerank_error_is_mapped_to_litellm_exception(respx_mock: respx.Mo
     assert "None - " not in str(exc_info.value)
 
 
+@pytest.mark.asyncio
+async def test_arerank_declared_authenticating_provider_skips_resolution(monkeypatch):
+    """Regression for the event-loop hazard in arerank's provider pre-resolution:
+    get_llm_provider runs the blocking OAuth device flow for github_copilot/chatgpt,
+    so arerank must adopt the declared provider instead of resolving it, while the
+    except path still maps with that declared provider."""
+    from litellm.llms.base_llm.chat.transformation import BaseLLMException
+
+    resolution_calls = []
+
+    def record_resolution(*args, **kwargs):
+        resolution_calls.append((args, kwargs))
+        return "gpt-4o", "github_copilot", None, None
+
+    def rerank_raises_provider_error(*args, **kwargs):
+        raise BaseLLMException(status_code=401, message='{"error":"bad key"}')
+
+    monkeypatch.setattr(litellm, "get_llm_provider", record_resolution)
+    monkeypatch.setattr("litellm.rerank_api.main.rerank", rerank_raises_provider_error)
+
+    with pytest.raises(litellm.AuthenticationError) as exc_info:
+        await litellm.arerank(
+            model="github_copilot/gpt-4o",
+            query=MARKER_QUERY,
+            documents=[MARKER_DOC],
+        )
+
+    assert resolution_calls == []
+    assert "Github_copilotException" in str(exc_info.value)
+    assert "None - " not in str(exc_info.value)
+
+
 @pytest.mark.asyncio
 async def test_together_rerank_async_honors_env_api_base(respx_mock: respx.MockRouter, monkeypatch):
     """Regression: TOGETHER_AI_API_BASE was honored by chat but ignored by rerank."""

From fcd9052179f039f075b3e25a8c1aec8657fc98a0 Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Tue, 1 Sep 2026 17:10:52 -0700
Subject: [PATCH 07/16] feat(proxy): honor model_info.display_name in the
 Anthropic-shaped /v1/models listing

---
 litellm/llms/anthropic/common_utils.py        | 17 ++--
 .../proxy/common_utils/model_listing_utils.py | 25 +++++-
 litellm/proxy/proxy_server.py                 | 21 +++--
 litellm/router.py                             | 20 +++++
 .../proxy/proxy_server/test_routes_models.py  | 78 +++++++++++++++++++
 .../test_team_model_name_translation.py       | 26 ++++++-
 tests/test_litellm/test_router.py             | 65 ++++++++++++++++
 7 files changed, 240 insertions(+), 12 deletions(-)

diff --git a/litellm/llms/anthropic/common_utils.py b/litellm/llms/anthropic/common_utils.py
index c60ebd844ba..d23690976ad 100644
--- a/litellm/llms/anthropic/common_utils.py
+++ b/litellm/llms/anthropic/common_utils.py
@@ -1378,31 +1378,38 @@ def process_anthropic_headers(headers: httpx.Headers | dict) -> dict:
     return additional_headers
 
 
-def _anthropic_model_entry(model: ModelInfoResponse, created_at: str) -> Mapping[str, object]:
+def _anthropic_model_entry(
+    model: ModelInfoResponse, created_at: str, display_names: Mapping[str, str]
+) -> Mapping[str, object]:
     return {  # mutable-ok: JSON response body, serialized by the route and never mutated
         "type": "model",
         "id": model["id"],
-        "display_name": model["id"],
+        "display_name": display_names.get(model["id"], model["id"]),
         "created_at": created_at,
         "max_input_tokens": model.get("max_input_tokens"),
         "max_tokens": model.get("max_output_tokens"),
     }
 
 
-def create_anthropic_model_list_response(models: Sequence[ModelInfoResponse]) -> Mapping[str, object]:
+def create_anthropic_model_list_response(
+    models: Sequence[ModelInfoResponse],
+    display_names: Mapping[str, str] = MappingProxyType({}),
+) -> Mapping[str, object]:
     """Build the Anthropic-native /v1/models envelope.
 
     Clients that send an anthropic-version header parse the Anthropic Models API
     shape (type/display_name/created_at plus has_more/first_id/last_id) and filter
     the list themselves, so every model is returned here. The token limits carry
     over from the OpenAI-shaped listing, named as the Messages API names them, and
-    are always present because the vendor shape declares them nullable, not optional
+    are always present because the vendor shape declares them nullable, not optional.
+    display_names maps a listed model id to a configured human-readable name; ids
+    without an entry fall back to the id itself, matching the vendor behavior
     """
     created_at: Final = (
         datetime.fromtimestamp(DEFAULT_MODEL_CREATED_AT_TIME, tz=timezone.utc).isoformat().replace("+00:00", "Z")
     )
     data: Final = [  # mutable-ok: JSON response body, serialized by the route and never mutated
-        _anthropic_model_entry(model, created_at) for model in models
+        _anthropic_model_entry(model, created_at, display_names) for model in models
     ]
     return {  # mutable-ok: JSON response body, serialized by the route and never mutated
         "data": data,
diff --git a/litellm/proxy/common_utils/model_listing_utils.py b/litellm/proxy/common_utils/model_listing_utils.py
index 9fd24162f7e..213a697b3dd 100644
--- a/litellm/proxy/common_utils/model_listing_utils.py
+++ b/litellm/proxy/common_utils/model_listing_utils.py
@@ -10,13 +10,36 @@ legacy internal names with `general_settings.use_team_public_model_name: false`.
 
 from __future__ import annotations
 
-from collections.abc import Mapping
+from collections.abc import Mapping, Sequence
+from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, cast
 
 if TYPE_CHECKING:
     from litellm.router import Router
 
 
+def configured_display_names(
+    entries: Sequence[tuple[str, str]],
+    llm_router: Router | None,
+) -> Mapping[str, str]:
+    """response_id -> configured `model_info.display_name` for the listing entries
+    that have one.
+
+    Metadata is looked up by each entry's internal lookup id (so team-scoped rows
+    resolve), while the returned map is keyed by the public response id the
+    Anthropic-shaped listing is built from. Entries without a configured name are
+    omitted so the listing falls back to the id itself.
+    """
+    if llm_router is None:
+        return MappingProxyType({})
+    resolved: Final = (
+        (response_id, llm_router.get_configured_display_name(lookup_id)) for response_id, lookup_id in entries
+    )
+    return MappingProxyType(
+        {response_id: display_name for response_id, display_name in resolved if display_name is not None}
+    )
+
+
 class TeamModelNameTranslator:
     """Translates internal team routing keys to their public names for the model
     listing/retrieve responses. Stateless; the live router and general_settings
diff --git a/litellm/proxy/proxy_server.py b/litellm/proxy/proxy_server.py
index 2c600667283..4f172ca29b9 100644
--- a/litellm/proxy/proxy_server.py
+++ b/litellm/proxy/proxy_server.py
@@ -351,7 +351,10 @@ from litellm.proxy.common_utils.load_config_utils import (
     get_file_contents_from_s3,
 )
 from litellm.proxy.common_utils.model_deprecation import collect_model_deprecations
-from litellm.proxy.common_utils.model_listing_utils import TeamModelNameTranslator
+from litellm.proxy.common_utils.model_listing_utils import (
+    TeamModelNameTranslator,
+    configured_display_names,
+)
 from litellm.proxy.common_utils.openai_endpoint_utils import (
     remove_sensitive_info_from_deployment,
 )
@@ -10193,7 +10196,8 @@ async def model_list(
         # The internal routing key drives the metadata/fallback lookup, while the
         # public name is what the client sees as the model id.
         model_data = []
-        for response_id, lookup_id in TeamModelNameTranslator.listing_entries(all_models, llm_router, settings):
+        admin_entries: Final = TeamModelNameTranslator.listing_entries(all_models, llm_router, settings)
+        for response_id, lookup_id in admin_entries:
             model_info = create_model_info_response(
                 model_id=lookup_id,
                 provider="openai",
@@ -10206,7 +10210,10 @@ async def model_list(
 
         if wants_anthropic_format:
             admin_listing: Final = cast(Sequence[ModelInfoResponse], model_data)  # cast-ok: rows built above
-            return create_anthropic_model_list_response(admin_listing)
+            return create_anthropic_model_list_response(
+                admin_listing,
+                display_names=configured_display_names(admin_entries, llm_router),
+            )
 
         return dict(
             data=model_data,
@@ -10237,7 +10244,8 @@ async def model_list(
     # The internal routing key drives the metadata/fallback lookup, while the
     # public name is what the client sees as the model id.
     model_data = []
-    for response_id, lookup_id in TeamModelNameTranslator.listing_entries(all_models, llm_router, settings):
+    entries: Final = TeamModelNameTranslator.listing_entries(all_models, llm_router, settings)
+    for response_id, lookup_id in entries:
         model_info = create_model_info_response(
             model_id=lookup_id,
             provider="openai",
@@ -10250,7 +10258,10 @@ async def model_list(
 
     if wants_anthropic_format:
         listing: Final = cast(Sequence[ModelInfoResponse], model_data)  # cast-ok: rows built above
-        return create_anthropic_model_list_response(listing)
+        return create_anthropic_model_list_response(
+            listing,
+            display_names=configured_display_names(entries, llm_router),
+        )
 
     return dict(
         data=model_data,
diff --git a/litellm/router.py b/litellm/router.py
index 471a1116f44..6245f8a6a03 100644
--- a/litellm/router.py
+++ b/litellm/router.py
@@ -9607,6 +9607,26 @@ class Router:
             coerce_token_limit(model_info.get("max_output_tokens")),
         )
 
+    def get_configured_display_name(self, model_name: str) -> "str | None":
+        """
+        Return the display_name explicitly configured in a concrete deployment's
+        model_info for model_name, via O(1) index lookup.
+
+        Returns None for wildcard-expanded or unknown names, and treats a
+        non-string or empty configured value as absent rather than failing the
+        listing. Like get_configured_token_limits, this never triggers pattern
+        matching or deep copies, so it is safe to call per listed model on the
+        /v1/models hot path.
+        """
+        deployment: Final = self.get_deployment_by_model_group_name(model_group_name=model_name)
+        if deployment is None:
+            return None
+
+        display_name: Final = deployment.model_info.get("display_name")
+        if isinstance(display_name, str) and display_name.strip():
+            return display_name
+        return None
+
     def get_deployment_credentials_with_provider(
         self, model_id: str, team_id: str | None = None
     ) -> dict[str, Any] | None:
diff --git a/tests/test_litellm/proxy/proxy_server/test_routes_models.py b/tests/test_litellm/proxy/proxy_server/test_routes_models.py
index 2b126b1ea95..bc6106a06f8 100644
--- a/tests/test_litellm/proxy/proxy_server/test_routes_models.py
+++ b/tests/test_litellm/proxy/proxy_server/test_routes_models.py
@@ -45,6 +45,7 @@ def patched_models(monkeypatch):
     deployment = MagicMock()
     deployment.litellm_params.model = "gpt-4"
     router.get_deployment_by_model_group_name = MagicMock(return_value=deployment)
+    router.get_configured_display_name = MagicMock(return_value=None)
 
     monkeypatch.setattr(proxy_server, "llm_router", router)
     monkeypatch.setattr(proxy_server, "prisma_client", MagicMock())
@@ -187,6 +188,83 @@ def test_anthropic_format_carries_router_configured_token_limits(client, auth_as
     assert (claude["max_input_tokens"], claude["max_tokens"]) == (500000, 4096)
 
 
+@pytest.mark.parametrize("path", ["/v1/models", "/models"])
+def test_anthropic_format_uses_configured_display_name(client, auth_as, patched_models, path):
+    """A deployment's ``model_info.display_name`` becomes the Anthropic-native
+    ``display_name`` so Claude Code's picker shows a clean name while the id keeps
+    routing; models without one keep the id fallback, and the OpenAI-shaped
+    listing carries no display_name either way."""
+
+    def _configured(model_name):
+        return "Kimi K3" if model_name == "gpt-4" else None
+
+    patched_models.get_configured_display_name = MagicMock(side_effect=_configured)
+
+    with auth_as():
+        anthropic_response = client.get(path, headers={"anthropic-version": "2023-06-01"})
+        openai_response = client.get(path)
+
+    assert anthropic_response.status_code == 200
+    gpt_4, claude = anthropic_response.json()["data"]
+    assert (gpt_4["id"], gpt_4["display_name"]) == ("gpt-4", "Kimi K3")
+    assert (claude["id"], claude["display_name"]) == ("claude-sonnet", "claude-sonnet")
+
+    assert openai_response.status_code == 200
+    openai_models = openai_response.json()["data"]
+    assert [m["id"] for m in openai_models] == ["gpt-4", "claude-sonnet"]
+    assert all("display_name" not in m for m in openai_models)
+
+
+@pytest.mark.parametrize("params", [{}, {"scope": "expand"}])
+def test_anthropic_display_name_resolved_via_internal_team_key(
+    client, auth_as, patched_models, monkeypatch, params
+):
+    """For a team-scoped row the configured display name must be looked up by the
+    internal routing key while the entry itself is keyed by the public name, so
+    the clean name lands on the id the client actually sees."""
+    from litellm.proxy import utils as proxy_utils
+    from litellm.proxy.auth import model_checks
+
+    internal_name = "model_name_team-1_c0ffee"
+
+    patched_models.get_model_list = MagicMock(
+        return_value=[
+            {
+                "model_name": internal_name,
+                "model_info": {
+                    "team_id": "team-1",
+                    "team_public_model_name": "gpt-4-team",
+                },
+            }
+        ]
+    )
+    patched_models.get_model_names = MagicMock(return_value=[internal_name])
+    patched_models.get_configured_display_name = MagicMock(
+        side_effect=lambda model_name: "Team GPT" if model_name == internal_name else None
+    )
+
+    async def _fake_get_available_models_for_user(**kwargs):
+        return [internal_name]
+
+    monkeypatch.setattr(
+        proxy_utils,
+        "get_available_models_for_user",
+        _fake_get_available_models_for_user,
+    )
+    monkeypatch.setattr(
+        model_checks, "get_complete_model_list", lambda **kwargs: [internal_name]
+    )
+
+    with auth_as():
+        response = client.get(
+            "/v1/models", params=params, headers={"anthropic-version": "2023-06-01"}
+        )
+
+    assert response.status_code == 200
+    (entry,) = response.json()["data"]
+    assert (entry["id"], entry["display_name"]) == ("gpt-4-team", "Team GPT")
+
+
 @pytest.mark.parametrize("path", ["/v1/models", "/models"])
 def test_get_models_invalid_scope_returns_400(client, auth_as, patched_models, path):
     """Pins: ``GET /v1/models``, ``GET /models`` (error path: invalid scope)."""
diff --git a/tests/test_litellm/proxy/proxy_server/test_team_model_name_translation.py b/tests/test_litellm/proxy/proxy_server/test_team_model_name_translation.py
index aa35fd64f18..0fb9b1a6d88 100644
--- a/tests/test_litellm/proxy/proxy_server/test_team_model_name_translation.py
+++ b/tests/test_litellm/proxy/proxy_server/test_team_model_name_translation.py
@@ -19,7 +19,10 @@ from litellm.proxy._types import (
     LitellmUserRoles,
     UserAPIKeyAuth,
 )
-from litellm.proxy.common_utils.model_listing_utils import TeamModelNameTranslator
+from litellm.proxy.common_utils.model_listing_utils import (
+    TeamModelNameTranslator,
+    configured_display_names,
+)
 from litellm.proxy.proxy_server import (
     _get_proxy_model_info,
     _translate_model_name_for_response,
@@ -1391,6 +1394,27 @@ def test_resolve_public_name_respects_legacy_flag():
     )
 
 
+def test_configured_display_names_keyed_by_response_id():
+    """The map is keyed by the public response id while the router lookup uses
+    the internal routing key, and entries without a configured name are omitted."""
+    router = MagicMock()
+    router.get_configured_display_name = MagicMock(
+        side_effect=lambda model_name: "Team Sonnet" if model_name == "model_name_team-abc-123_4a6b8" else None
+    )
+
+    assert configured_display_names(
+        entries=[
+            ("team-claude-sonnet", "model_name_team-abc-123_4a6b8"),
+            ("gpt-4o", "gpt-4o"),
+        ],
+        llm_router=router,
+    ) == {"team-claude-sonnet": "Team Sonnet"}
+
+
+def test_configured_display_names_empty_without_router():
+    assert configured_display_names(entries=[("gpt-4o", "gpt-4o")], llm_router=None) == {}
+
+
 @pytest.mark.asyncio
 async def test_retrieve_model_by_public_name_returns_200(monkeypatch):
     """Regression: `GET /v1/models/{public_name}` must NOT 404. The listing
diff --git a/tests/test_litellm/test_router.py b/tests/test_litellm/test_router.py
index 44c1cdbff06..f4ea9b03a80 100644
--- a/tests/test_litellm/test_router.py
+++ b/tests/test_litellm/test_router.py
@@ -7271,6 +7271,71 @@ def test_get_configured_token_limits_coerces_numeric_strings():
     assert router.get_configured_token_limits("quoted-limits-model") == (32000, 8000)
 
 
+def test_get_configured_display_name_reads_deployment_model_info():
+    router = litellm.Router(
+        model_list=[
+            {
+                "model_name": "Kimi K3-claude-compatible",
+                "litellm_params": {"model": "openai/some-unmapped-model"},
+                "model_info": {"display_name": "Kimi K3"},
+            }
+        ]
+    )
+
+    assert router.get_configured_display_name("Kimi K3-claude-compatible") == "Kimi K3"
+
+
+def test_get_configured_display_name_returns_none_for_unset_or_unknown():
+    router = litellm.Router(
+        model_list=[
+            {
+                "model_name": "no-display-model",
+                "litellm_params": {"model": "openai/some-unmapped-model"},
+            }
+        ]
+    )
+
+    assert router.get_configured_display_name("no-display-model") is None
+    assert router.get_configured_display_name("not-a-real-model") is None
+
+
+def test_get_configured_display_name_skips_wildcard_pattern_matching():
+    router = litellm.Router(
+        model_list=[
+            {
+                "model_name": "bedrock/*",
+                "litellm_params": {"model": "bedrock/*"},
+                "model_info": {"display_name": "Bedrock"},
+            }
+        ]
+    )
+
+    with patch.object(
+        router.pattern_router, "route", side_effect=AssertionError("pattern route called")
+    ):
+        assert (
+            router.get_configured_display_name("bedrock/anthropic.claude-3-5-sonnet-20240620-v1:0")
+            is None
+        )
+
+
+def test_get_configured_display_name_treats_malformed_values_as_absent():
+    malformed = ["", "   ", 12345, ["Kimi K3"], {"name": "Kimi K3"}, True]
+    router = litellm.Router(
+        model_list=[
+            {
+                "model_name": f"bad-display-{i}",
+                "litellm_params": {"model": "openai/some-unmapped-model"},
+                "model_info": {"display_name": bad},
+            }
+            for i, bad in enumerate(malformed)
+        ]
+    )
+
+    for i in range(len(malformed)):
+        assert router.get_configured_display_name(f"bad-display-{i}") is None
+
+
 @pytest.mark.asyncio
 async def test_acreate_batch_disable_fallbacks_surfaces_owning_provider_error():
     router = litellm.Router(

From f49a3e15a8b936c94e5d050017f6887e8d34e997 Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Tue, 1 Sep 2026 19:12:07 -0700
Subject: [PATCH 08/16] test(e2e): read JUnit properties off the real collected
 pytest Item

tests/e2e/test_junit_properties.py fed a hand-rolled FakeItem to
result_properties and attach_result_properties, both typed pytest.Item,
so uv run basedpyright tests/e2e reported 3 reportArgumentType errors on
litellm_internal_staging and every make check that scopes a litellm/ or
tests/e2e/ Python file failed.

Each test now looks up its own collected Item in request.session.items
and applies the covers marker at run time through request.applymarker,
so the coverage registry's collect-only pass never sees the test ids and
the production functions keep their pytest.Item signatures. No casts, no
ignores.

Resolves LIT-6669
---
 tests/e2e/test_junit_properties.py | 45 ++++++++++--------------------
 1 file changed, 15 insertions(+), 30 deletions(-)

diff --git a/tests/e2e/test_junit_properties.py b/tests/e2e/test_junit_properties.py
index c0596177cc1..02c1413c840 100644
--- a/tests/e2e/test_junit_properties.py
+++ b/tests/e2e/test_junit_properties.py
@@ -24,25 +24,10 @@ from junit_properties import (
 )
 
 
-class FakeMarker:
-    def __init__(self, name: str, *args: object) -> None:
-        self.name = name
-        self.args = args
-
-
-class FakeItem:
-    """The three attributes junit_properties reads off a pytest Item."""
-
-    def __init__(
-        self, nodeid: str, location: tuple[str, int | None, str], markers: tuple[FakeMarker, ...] = ()
-    ) -> None:
-        self.nodeid = nodeid
-        self.location = location
-        self.user_properties: list[tuple[str, str]] = []
-        self._markers = markers
-
-    def iter_markers(self, name: str):
-        return (marker for marker in self._markers if marker.name == name)
+def collected_item(request: pytest.FixtureRequest, name: str) -> pytest.Item:
+    """The Item pytest collected for test ``name`` in this file: the real nodeid,
+    location and marker machinery the collection hook reads, as pytest built it."""
+    return next(item for item in request.session.items if item.path == request.path and item.name == name)
 
 
 def repo_root() -> Path | None:
@@ -109,22 +94,22 @@ class TestSourceFromLocation:
 
 
 class TestResultProperties:
-    def test_every_test_carries_package_covers_and_source(self) -> None:
-        item = FakeItem(
-            "logging/test_x.py::TestFoo::test_bar",
-            ("logging/test_x.py", 40, "TestFoo.test_bar"),
-            (FakeMarker("covers", "LOG-1", "LOG-2"),),
-        )
-        assert result_properties(item) == (
-            ("package", "logging"),
+    def test_every_test_carries_package_covers_and_source(self, request: pytest.FixtureRequest) -> None:
+        """Read off this test's own collected Item, so the nodeid and location are
+        whatever pytest reports for the launch shape in use, and the marker is added
+        at run time so the coverage registry's collect-only pass never sees it."""
+        test = type(self).test_every_test_carries_package_covers_and_source
+        request.applymarker(pytest.mark.covers("LOG-1", "LOG-2"))
+        assert result_properties(collected_item(request, test.__name__)) == (
+            ("package", "root"),
             ("covers", "LOG-1,LOG-2"),
-            ("source", "tests/e2e/logging/test_x.py:41"),
+            ("source", f"tests/e2e/test_junit_properties.py:{test.__code__.co_firstlineno}"),
         )
 
-    def test_attach_is_idempotent(self) -> None:
+    def test_attach_is_idempotent(self, request: pytest.FixtureRequest) -> None:
         """Collection can run the hook more than once; a second pass must not
         double the <property> entries in the report."""
-        item = FakeItem("logging/test_x.py::test_bar", ("logging/test_x.py", 40, "test_bar"))
+        item = collected_item(request, type(self).test_attach_is_idempotent.__name__)
         attach_result_properties(item)
         attach_result_properties(item)
         assert [name for name, _ in item.user_properties] == ["package", "covers", "source"]

From a7836ede15bb4f62d8f44bdb991402a9829727e3 Mon Sep 17 00:00:00 2001
From: mateo <mateo@berri.ai>
Date: Wed, 2 Sep 2026 14:48:51 +0000
Subject: [PATCH 09/16] fix(models): absorb open registry PRs: govcloud bedrock
 and mantle, azure gov, openai tiered long-context, scaleway, together
 qwen3.8, azure ai cache and kimi k2.7 code, azure mai deprecations

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 ...odel_prices_and_context_window_backup.json | 605 +++++++++++++++++-
 model_prices_and_context_window.json          | 605 +++++++++++++++++-
 .../llm_cost_calc/test_llm_cost_calc_utils.py |   6 +-
 ...penai_service_tier_long_context_pricing.py | 156 +++++
 whitelisted_bedrock_models.txt                |  14 +
 5 files changed, 1341 insertions(+), 45 deletions(-)
 create mode 100644 tests/test_litellm/test_openai_service_tier_long_context_pricing.py

diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index c710db1a749..87d348f4752 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -9643,7 +9643,8 @@
         "supported_endpoints": [
             "/v1/images/generations",
             "/v1/images/edits"
-        ]
+        ],
+        "deprecation_date": "2026-10-01"
     },
     "azure_ai/MAI-Image-2.5-Flash": {
         "input_cost_per_image_token": 1.75e-06,
@@ -9656,7 +9657,8 @@
         "supported_endpoints": [
             "/v1/images/generations",
             "/v1/images/edits"
-        ]
+        ],
+        "deprecation_date": "2026-10-01"
     },
     "azure_ai/MAI-Image-2e": {
         "deprecation_date": "2026-08-15",
@@ -10155,7 +10157,9 @@
         "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/deepseek/",
         "supports_function_calling": true,
         "supports_reasoning": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "cache_read_input_token_cost": 1.45e-07,
+        "supports_prompt_caching": true
     },
     "azure_ai/deepseek-v4-flash": {
         "deprecation_date": "2028-02-20",
@@ -10169,18 +10173,20 @@
         "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/deepseek/",
         "supports_function_calling": true,
         "supports_reasoning": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "cache_read_input_token_cost": 2.8e-08,
+        "supports_prompt_caching": true
     },
     "azure_ai/deepseek-v4-flash-0731": {
-        "cache_read_input_token_cost": 2.8e-08,
+        "cache_read_input_token_cost": 1.4e-08,
         "deprecation_date": "2026-12-03",
-        "input_cost_per_token": 1.9e-07,
+        "input_cost_per_token": 4.4e-07,
         "litellm_provider": "azure_ai",
         "max_input_tokens": 1000000,
         "max_output_tokens": 128000,
         "max_tokens": 128000,
         "mode": "chat",
-        "output_cost_per_token": 5.1e-07,
+        "output_cost_per_token": 1.32e-06,
         "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/deepseek/",
         "supports_function_calling": true,
         "supports_prompt_caching": true,
@@ -10400,11 +10406,13 @@
         "max_tokens": 262144,
         "mode": "chat",
         "output_cost_per_token": 3e-06,
-        "source": "https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/kimi-k2-5-now-in-microsoft-foundry/4492321",
+        "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/kimi/",
         "supports_function_calling": true,
         "supports_tool_choice": true,
         "supports_video_input": true,
-        "supports_vision": true
+        "supports_vision": true,
+        "cache_read_input_token_cost": 1e-07,
+        "supports_prompt_caching": true
     },
     "azure_ai/kimi-k2.6": {
         "deprecation_date": "2027-04-16",
@@ -10415,7 +10423,7 @@
         "max_tokens": 262144,
         "mode": "chat",
         "output_cost_per_token": 4e-06,
-        "source": "https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/introducing-kimi-k2-6-in-microsoft-foundry/4513125",
+        "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/kimi/",
         "supported_modalities": [
             "text",
             "image"
@@ -10426,7 +10434,9 @@
         "supports_function_calling": true,
         "supports_reasoning": true,
         "supports_tool_choice": true,
-        "supports_vision": true
+        "supports_vision": true,
+        "cache_read_input_token_cost": 1.6e-07,
+        "supports_prompt_caching": true
     },
     "azure_ai/ministral-3b": {
         "input_cost_per_token": 4e-08,
@@ -12110,7 +12120,7 @@
         "max_output_tokens": 2048,
         "max_tokens": 2048,
         "mode": "chat",
-        "output_cost_per_token": 2.65e-06,
+        "output_cost_per_token": 6e-07,
         "supports_pdf_input": true
     },
     "bedrock/us-west-1/meta.llama3-70b-instruct-v1:0": {
@@ -29098,16 +29108,19 @@
         "cache_creation_input_token_cost": 5e-06,
         "cache_creation_input_token_cost_above_272k_tokens": 1e-05,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 5e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 2e-05,
         "cache_creation_input_token_cost_flex": 2.5e-06,
         "cache_creation_input_token_cost_priority": 1e-05,
         "cache_read_input_token_cost": 4e-07,
         "cache_read_input_token_cost_above_272k_tokens": 8e-07,
         "cache_read_input_token_cost_above_272k_tokens_flex": 4e-07,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 1.6e-06,
         "cache_read_input_token_cost_flex": 2e-07,
         "cache_read_input_token_cost_priority": 8e-07,
         "input_cost_per_token": 4e-06,
         "input_cost_per_token_above_272k_tokens": 8e-06,
         "input_cost_per_token_above_272k_tokens_flex": 4e-06,
+        "input_cost_per_token_above_272k_tokens_priority": 1.6e-05,
         "input_cost_per_token_batches": 2e-06,
         "input_cost_per_token_flex": 2e-06,
         "input_cost_per_token_priority": 8e-06,
@@ -29119,6 +29132,7 @@
         "output_cost_per_token": 2e-05,
         "output_cost_per_token_above_272k_tokens": 3e-05,
         "output_cost_per_token_above_272k_tokens_flex": 1.5e-05,
+        "output_cost_per_token_above_272k_tokens_priority": 6e-05,
         "output_cost_per_token_batches": 1e-05,
         "output_cost_per_token_flex": 1e-05,
         "output_cost_per_token_priority": 4e-05,
@@ -29161,16 +29175,19 @@
         "cache_creation_input_token_cost": 5e-06,
         "cache_creation_input_token_cost_above_272k_tokens": 1e-05,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 5e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 2e-05,
         "cache_creation_input_token_cost_flex": 2.5e-06,
         "cache_creation_input_token_cost_priority": 1e-05,
         "cache_read_input_token_cost": 4e-07,
         "cache_read_input_token_cost_above_272k_tokens": 8e-07,
         "cache_read_input_token_cost_above_272k_tokens_flex": 4e-07,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 1.6e-06,
         "cache_read_input_token_cost_flex": 2e-07,
         "cache_read_input_token_cost_priority": 8e-07,
         "input_cost_per_token": 4e-06,
         "input_cost_per_token_above_272k_tokens": 8e-06,
         "input_cost_per_token_above_272k_tokens_flex": 4e-06,
+        "input_cost_per_token_above_272k_tokens_priority": 1.6e-05,
         "input_cost_per_token_batches": 2e-06,
         "input_cost_per_token_flex": 2e-06,
         "input_cost_per_token_priority": 8e-06,
@@ -29182,6 +29199,7 @@
         "output_cost_per_token": 2e-05,
         "output_cost_per_token_above_272k_tokens": 3e-05,
         "output_cost_per_token_above_272k_tokens_flex": 1.5e-05,
+        "output_cost_per_token_above_272k_tokens_priority": 6e-05,
         "output_cost_per_token_batches": 1e-05,
         "output_cost_per_token_flex": 1e-05,
         "output_cost_per_token_priority": 4e-05,
@@ -29225,16 +29243,19 @@
         "cache_creation_input_token_cost": 2.5e-06,
         "cache_creation_input_token_cost_above_272k_tokens": 5e-06,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 2.5e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 1e-05,
         "cache_creation_input_token_cost_flex": 1.25e-06,
         "cache_creation_input_token_cost_priority": 5e-06,
         "cache_read_input_token_cost": 2e-07,
         "cache_read_input_token_cost_above_272k_tokens": 4e-07,
         "cache_read_input_token_cost_above_272k_tokens_flex": 2e-07,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 8e-07,
         "cache_read_input_token_cost_flex": 1e-07,
         "cache_read_input_token_cost_priority": 4e-07,
         "input_cost_per_token": 2e-06,
         "input_cost_per_token_above_272k_tokens": 4e-06,
         "input_cost_per_token_above_272k_tokens_flex": 2e-06,
+        "input_cost_per_token_above_272k_tokens_priority": 8e-06,
         "input_cost_per_token_batches": 1e-06,
         "input_cost_per_token_flex": 1e-06,
         "input_cost_per_token_priority": 4e-06,
@@ -29246,6 +29267,7 @@
         "output_cost_per_token": 1.2e-05,
         "output_cost_per_token_above_272k_tokens": 1.8e-05,
         "output_cost_per_token_above_272k_tokens_flex": 9e-06,
+        "output_cost_per_token_above_272k_tokens_priority": 3.6e-05,
         "output_cost_per_token_batches": 6e-06,
         "output_cost_per_token_flex": 6e-06,
         "output_cost_per_token_priority": 2.4e-05,
@@ -29288,16 +29310,19 @@
         "cache_creation_input_token_cost": 2.5e-07,
         "cache_creation_input_token_cost_above_272k_tokens": 5e-07,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 2.5e-07,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 1e-06,
         "cache_creation_input_token_cost_flex": 1.25e-07,
         "cache_creation_input_token_cost_priority": 5e-07,
         "cache_read_input_token_cost": 2e-08,
         "cache_read_input_token_cost_above_272k_tokens": 4e-08,
         "cache_read_input_token_cost_above_272k_tokens_flex": 2e-08,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 8e-08,
         "cache_read_input_token_cost_flex": 1e-08,
         "cache_read_input_token_cost_priority": 4e-08,
         "input_cost_per_token": 2e-07,
         "input_cost_per_token_above_272k_tokens": 4e-07,
         "input_cost_per_token_above_272k_tokens_flex": 2e-07,
+        "input_cost_per_token_above_272k_tokens_priority": 8e-07,
         "input_cost_per_token_batches": 1e-07,
         "input_cost_per_token_flex": 1e-07,
         "input_cost_per_token_priority": 4e-07,
@@ -29309,6 +29334,7 @@
         "output_cost_per_token": 1.2e-06,
         "output_cost_per_token_above_272k_tokens": 1.8e-06,
         "output_cost_per_token_above_272k_tokens_flex": 9e-07,
+        "output_cost_per_token_above_272k_tokens_priority": 3.6e-06,
         "output_cost_per_token_batches": 6e-07,
         "output_cost_per_token_flex": 6e-07,
         "output_cost_per_token_priority": 2.4e-06,
@@ -29548,7 +29574,10 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": true,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": false
+        "supports_minimal_reasoning_effort": false,
+        "input_cost_per_token_above_272k_tokens_flex": 5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 2.25e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 5e-07
     },
     "gpt-5.5-2026-04-23": {
         "cache_read_input_token_cost": 5e-07,
@@ -29602,7 +29631,10 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": true,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": false
+        "supports_minimal_reasoning_effort": false,
+        "input_cost_per_token_above_272k_tokens_flex": 5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 2.25e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 5e-07
     },
     "gpt-5.5-pro": {
         "cache_read_input_token_cost": 3e-06,
@@ -29751,7 +29783,10 @@
         "supports_none_reasoning_effort": true,
         "default_reasoning_effort": "none",
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 2.5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 1.125e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 2.5e-07
     },
     "gpt-5.4-2026-03-05": {
         "cache_read_input_token_cost": 2.5e-07,
@@ -29800,7 +29835,10 @@
         "supports_none_reasoning_effort": true,
         "default_reasoning_effort": "none",
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 2.5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 1.125e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 2.5e-07
     },
     "gpt-5.4-pro": {
         "cache_read_input_token_cost": 3e-06,
@@ -29849,7 +29887,9 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": false,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 3e-05,
+        "output_cost_per_token_above_272k_tokens_flex": 0.000135
     },
     "gpt-5.4-pro-2026-03-05": {
         "cache_read_input_token_cost": 3e-06,
@@ -29898,7 +29938,9 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": false,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 3e-05,
+        "output_cost_per_token_above_272k_tokens_flex": 0.000135
     },
     "gpt-5.4-mini": {
         "cache_read_input_token_cost": 7.5e-08,
@@ -41350,13 +41392,13 @@
         "source": "https://docs.together.ai/docs/serverless-models"
     },
     "together_ai/Qwen/Qwen3.8-2.4T-A95B": {
-        "cache_read_input_token_cost": 5e-07,
-        "input_cost_per_token": 2.5e-06,
+        "cache_read_input_token_cost": 2.5e-07,
+        "input_cost_per_token": 2e-06,
         "litellm_provider": "together_ai",
         "max_input_tokens": 1010000,
         "max_tokens": 1010000,
         "mode": "chat",
-        "output_cost_per_token": 6.25e-06,
+        "output_cost_per_token": 6e-06,
         "source": "https://docs.together.ai/docs/serverless-models",
         "supports_prompt_caching": true
     },
@@ -57558,5 +57600,526 @@
         "supported_endpoints": [
             "/v1/audio/transcriptions"
         ]
+    },
+    "scaleway/glm-5.2": {
+        "input_cost_per_token": 1.8e-06,
+        "litellm_provider": "scaleway",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 16384,
+        "max_tokens": 16384,
+        "mode": "chat",
+        "output_cost_per_token": 5.5e-06,
+        "source": "https://www.scaleway.com/en/pricing/model-as-a-service/",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_vision": false
+    },
+    "scaleway/deepseek-v4-flash-0731": {
+        "cache_read_input_token_cost": 8e-08,
+        "input_cost_per_token": 4e-07,
+        "litellm_provider": "scaleway",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 32768,
+        "max_tokens": 32768,
+        "mode": "chat",
+        "output_cost_per_token": 8e-07,
+        "source": "https://www.scaleway.com/en/pricing/model-as-a-service/",
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_vision": false
+    },
+    "azure_ai/kimi-k2.7-code": {
+        "cache_read_input_token_cost": 1.9e-07,
+        "deprecation_date": "2026-10-03",
+        "input_cost_per_token": 9.5e-07,
+        "litellm_provider": "azure_ai",
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
+        "mode": "chat",
+        "output_cost_per_token": 4e-06,
+        "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/kimi/",
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
+    "bedrock/us-gov-west-1/nvidia.nemotron-nano-3-30b": {
+        "input_cost_per_token": 7.2e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 262144,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 2.88e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_native_structured_output": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/nvidia.nemotron-nano-12b-v2": {
+        "input_cost_per_token": 2.4e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_system_messages": true,
+        "supports_vision": true
+    },
+    "bedrock/us-gov-west-1/nvidia.nemotron-super-3-120b": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 32768,
+        "max_tokens": 32768,
+        "mode": "chat",
+        "output_cost_per_token": 7.8e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/openai.gpt-oss-20b-1:0": {
+        "input_cost_per_token": 8.4e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3.6e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/openai.gpt-oss-120b-1:0": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/anthropic.claude-sonnet-5": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 3e-06,
+        "cache_creation_input_token_cost_above_1hr": 4.8e-06,
+        "cache_read_input_token_cost": 2.4e-07,
+        "input_cost_per_token": 2.4e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": false,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock/us-gov-west-1/anthropic.claude-opus-4-8": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 7.5e-06,
+        "cache_creation_input_token_cost_above_1hr": 1.2e-05,
+        "cache_read_input_token_cost": 6e-07,
+        "input_cost_per_token": 6e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": true,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock/us-gov-east-1/nvidia.nemotron-nano-3-30b": {
+        "input_cost_per_token": 7.2e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 262144,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 2.88e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_native_structured_output": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/nvidia.nemotron-nano-12b-v2": {
+        "input_cost_per_token": 2.4e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_system_messages": true,
+        "supports_vision": true
+    },
+    "bedrock/us-gov-east-1/nvidia.nemotron-super-3-120b": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 32768,
+        "max_tokens": 32768,
+        "mode": "chat",
+        "output_cost_per_token": 7.8e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/openai.gpt-oss-20b-1:0": {
+        "input_cost_per_token": 8.4e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3.6e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/openai.gpt-oss-120b-1:0": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/anthropic.claude-sonnet-5": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 3e-06,
+        "cache_creation_input_token_cost_above_1hr": 4.8e-06,
+        "cache_read_input_token_cost": 2.4e-07,
+        "input_cost_per_token": 2.4e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": false,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock/us-gov-east-1/anthropic.claude-opus-4-8": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 7.5e-06,
+        "cache_creation_input_token_cost_above_1hr": 1.2e-05,
+        "cache_read_input_token_cost": 6e-07,
+        "input_cost_per_token": 6e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": true,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock_mantle/us-gov-west-1/openai.gpt-5.6-terra": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 2.64e-06,
+        "input_cost_per_token_above_272k_tokens": 5.28e-06,
+        "cache_creation_input_token_cost": 3.3e-06,
+        "cache_creation_input_token_cost_above_272k_tokens": 6.6e-06,
+        "cache_read_input_token_cost": 2.64e-07,
+        "cache_read_input_token_cost_above_272k_tokens": 5.28e-07,
+        "output_cost_per_token": 1.584e-05,
+        "output_cost_per_token_above_272k_tokens": 2.376e-05
+    },
+    "bedrock_mantle/us-gov-west-1/openai.gpt-5.6-luna": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 2.64e-07,
+        "input_cost_per_token_above_272k_tokens": 5.28e-07,
+        "cache_creation_input_token_cost": 3.3e-07,
+        "cache_creation_input_token_cost_above_272k_tokens": 6.6e-07,
+        "cache_read_input_token_cost": 2.64e-08,
+        "cache_read_input_token_cost_above_272k_tokens": 5.28e-08,
+        "output_cost_per_token": 1.584e-06,
+        "output_cost_per_token_above_272k_tokens": 2.376e-06
+    },
+    "bedrock_mantle/us-gov-west-1/openai.gpt-5.4": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 3.3e-06,
+        "cache_read_input_token_cost": 3.3e-07,
+        "output_cost_per_token": 1.98e-05
+    },
+    "bedrock_mantle/us-gov-west-1/xai.grok-4.3": {
+        "use_openai_responses_path": true,
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 131072,
+        "max_output_tokens": 16384,
+        "max_tokens": 16384,
+        "mode": "chat",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/responses"
+        ],
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "input_cost_per_token": 1.5e-06,
+        "output_cost_per_token": 3e-06,
+        "cache_read_input_token_cost": 2.4e-07
+    },
+    "bedrock_mantle/us-gov-east-1/openai.gpt-5.4": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 3.3e-06,
+        "cache_read_input_token_cost": 3.3e-07,
+        "output_cost_per_token": 1.98e-05
+    },
+    "azure/us-gov/gpt-5.1": {
+        "cache_read_input_token_cost": 1.71875e-07,
+        "default_reasoning_effort": "none",
+        "deprecation_date": "2027-05-15",
+        "input_cost_per_token": 1.71875e-06,
+        "litellm_provider": "azure",
+        "max_input_tokens": 272000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.375e-05,
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/batch",
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text",
+            "image"
+        ],
+        "supports_function_calling": true,
+        "supports_native_streaming": true,
+        "supports_none_reasoning_effort": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
+    "azure/us-gov/o3-mini": {
+        "cache_read_input_token_cost": 7.57e-07,
+        "deprecation_date": "2026-10-01",
+        "input_cost_per_token": 1.513e-06,
+        "litellm_provider": "azure",
+        "max_input_tokens": 200000,
+        "max_output_tokens": 100000,
+        "max_tokens": 100000,
+        "mode": "chat",
+        "output_cost_per_token": 6.05e-06,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": false
+    },
+    "azure/us-gov/text-embedding-3-large": {
+        "deprecation_date": "2028-02-09",
+        "input_cost_per_token": 1.63e-07,
+        "litellm_provider": "azure",
+        "max_input_tokens": 8191,
+        "max_tokens": 8191,
+        "mode": "embedding",
+        "output_cost_per_token": 0.0
+    },
+    "azure/us-gov/text-embedding-3-small": {
+        "deprecation_date": "2028-02-09",
+        "input_cost_per_token": 2.5e-08,
+        "litellm_provider": "azure",
+        "max_input_tokens": 8191,
+        "max_tokens": 8191,
+        "mode": "embedding",
+        "output_cost_per_token": 0.0
     }
 }
diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json
index c710db1a749..87d348f4752 100644
--- a/model_prices_and_context_window.json
+++ b/model_prices_and_context_window.json
@@ -9643,7 +9643,8 @@
         "supported_endpoints": [
             "/v1/images/generations",
             "/v1/images/edits"
-        ]
+        ],
+        "deprecation_date": "2026-10-01"
     },
     "azure_ai/MAI-Image-2.5-Flash": {
         "input_cost_per_image_token": 1.75e-06,
@@ -9656,7 +9657,8 @@
         "supported_endpoints": [
             "/v1/images/generations",
             "/v1/images/edits"
-        ]
+        ],
+        "deprecation_date": "2026-10-01"
     },
     "azure_ai/MAI-Image-2e": {
         "deprecation_date": "2026-08-15",
@@ -10155,7 +10157,9 @@
         "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/deepseek/",
         "supports_function_calling": true,
         "supports_reasoning": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "cache_read_input_token_cost": 1.45e-07,
+        "supports_prompt_caching": true
     },
     "azure_ai/deepseek-v4-flash": {
         "deprecation_date": "2028-02-20",
@@ -10169,18 +10173,20 @@
         "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/deepseek/",
         "supports_function_calling": true,
         "supports_reasoning": true,
-        "supports_tool_choice": true
+        "supports_tool_choice": true,
+        "cache_read_input_token_cost": 2.8e-08,
+        "supports_prompt_caching": true
     },
     "azure_ai/deepseek-v4-flash-0731": {
-        "cache_read_input_token_cost": 2.8e-08,
+        "cache_read_input_token_cost": 1.4e-08,
         "deprecation_date": "2026-12-03",
-        "input_cost_per_token": 1.9e-07,
+        "input_cost_per_token": 4.4e-07,
         "litellm_provider": "azure_ai",
         "max_input_tokens": 1000000,
         "max_output_tokens": 128000,
         "max_tokens": 128000,
         "mode": "chat",
-        "output_cost_per_token": 5.1e-07,
+        "output_cost_per_token": 1.32e-06,
         "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/deepseek/",
         "supports_function_calling": true,
         "supports_prompt_caching": true,
@@ -10400,11 +10406,13 @@
         "max_tokens": 262144,
         "mode": "chat",
         "output_cost_per_token": 3e-06,
-        "source": "https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/kimi-k2-5-now-in-microsoft-foundry/4492321",
+        "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/kimi/",
         "supports_function_calling": true,
         "supports_tool_choice": true,
         "supports_video_input": true,
-        "supports_vision": true
+        "supports_vision": true,
+        "cache_read_input_token_cost": 1e-07,
+        "supports_prompt_caching": true
     },
     "azure_ai/kimi-k2.6": {
         "deprecation_date": "2027-04-16",
@@ -10415,7 +10423,7 @@
         "max_tokens": 262144,
         "mode": "chat",
         "output_cost_per_token": 4e-06,
-        "source": "https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/introducing-kimi-k2-6-in-microsoft-foundry/4513125",
+        "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/kimi/",
         "supported_modalities": [
             "text",
             "image"
@@ -10426,7 +10434,9 @@
         "supports_function_calling": true,
         "supports_reasoning": true,
         "supports_tool_choice": true,
-        "supports_vision": true
+        "supports_vision": true,
+        "cache_read_input_token_cost": 1.6e-07,
+        "supports_prompt_caching": true
     },
     "azure_ai/ministral-3b": {
         "input_cost_per_token": 4e-08,
@@ -12110,7 +12120,7 @@
         "max_output_tokens": 2048,
         "max_tokens": 2048,
         "mode": "chat",
-        "output_cost_per_token": 2.65e-06,
+        "output_cost_per_token": 6e-07,
         "supports_pdf_input": true
     },
     "bedrock/us-west-1/meta.llama3-70b-instruct-v1:0": {
@@ -29098,16 +29108,19 @@
         "cache_creation_input_token_cost": 5e-06,
         "cache_creation_input_token_cost_above_272k_tokens": 1e-05,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 5e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 2e-05,
         "cache_creation_input_token_cost_flex": 2.5e-06,
         "cache_creation_input_token_cost_priority": 1e-05,
         "cache_read_input_token_cost": 4e-07,
         "cache_read_input_token_cost_above_272k_tokens": 8e-07,
         "cache_read_input_token_cost_above_272k_tokens_flex": 4e-07,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 1.6e-06,
         "cache_read_input_token_cost_flex": 2e-07,
         "cache_read_input_token_cost_priority": 8e-07,
         "input_cost_per_token": 4e-06,
         "input_cost_per_token_above_272k_tokens": 8e-06,
         "input_cost_per_token_above_272k_tokens_flex": 4e-06,
+        "input_cost_per_token_above_272k_tokens_priority": 1.6e-05,
         "input_cost_per_token_batches": 2e-06,
         "input_cost_per_token_flex": 2e-06,
         "input_cost_per_token_priority": 8e-06,
@@ -29119,6 +29132,7 @@
         "output_cost_per_token": 2e-05,
         "output_cost_per_token_above_272k_tokens": 3e-05,
         "output_cost_per_token_above_272k_tokens_flex": 1.5e-05,
+        "output_cost_per_token_above_272k_tokens_priority": 6e-05,
         "output_cost_per_token_batches": 1e-05,
         "output_cost_per_token_flex": 1e-05,
         "output_cost_per_token_priority": 4e-05,
@@ -29161,16 +29175,19 @@
         "cache_creation_input_token_cost": 5e-06,
         "cache_creation_input_token_cost_above_272k_tokens": 1e-05,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 5e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 2e-05,
         "cache_creation_input_token_cost_flex": 2.5e-06,
         "cache_creation_input_token_cost_priority": 1e-05,
         "cache_read_input_token_cost": 4e-07,
         "cache_read_input_token_cost_above_272k_tokens": 8e-07,
         "cache_read_input_token_cost_above_272k_tokens_flex": 4e-07,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 1.6e-06,
         "cache_read_input_token_cost_flex": 2e-07,
         "cache_read_input_token_cost_priority": 8e-07,
         "input_cost_per_token": 4e-06,
         "input_cost_per_token_above_272k_tokens": 8e-06,
         "input_cost_per_token_above_272k_tokens_flex": 4e-06,
+        "input_cost_per_token_above_272k_tokens_priority": 1.6e-05,
         "input_cost_per_token_batches": 2e-06,
         "input_cost_per_token_flex": 2e-06,
         "input_cost_per_token_priority": 8e-06,
@@ -29182,6 +29199,7 @@
         "output_cost_per_token": 2e-05,
         "output_cost_per_token_above_272k_tokens": 3e-05,
         "output_cost_per_token_above_272k_tokens_flex": 1.5e-05,
+        "output_cost_per_token_above_272k_tokens_priority": 6e-05,
         "output_cost_per_token_batches": 1e-05,
         "output_cost_per_token_flex": 1e-05,
         "output_cost_per_token_priority": 4e-05,
@@ -29225,16 +29243,19 @@
         "cache_creation_input_token_cost": 2.5e-06,
         "cache_creation_input_token_cost_above_272k_tokens": 5e-06,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 2.5e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 1e-05,
         "cache_creation_input_token_cost_flex": 1.25e-06,
         "cache_creation_input_token_cost_priority": 5e-06,
         "cache_read_input_token_cost": 2e-07,
         "cache_read_input_token_cost_above_272k_tokens": 4e-07,
         "cache_read_input_token_cost_above_272k_tokens_flex": 2e-07,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 8e-07,
         "cache_read_input_token_cost_flex": 1e-07,
         "cache_read_input_token_cost_priority": 4e-07,
         "input_cost_per_token": 2e-06,
         "input_cost_per_token_above_272k_tokens": 4e-06,
         "input_cost_per_token_above_272k_tokens_flex": 2e-06,
+        "input_cost_per_token_above_272k_tokens_priority": 8e-06,
         "input_cost_per_token_batches": 1e-06,
         "input_cost_per_token_flex": 1e-06,
         "input_cost_per_token_priority": 4e-06,
@@ -29246,6 +29267,7 @@
         "output_cost_per_token": 1.2e-05,
         "output_cost_per_token_above_272k_tokens": 1.8e-05,
         "output_cost_per_token_above_272k_tokens_flex": 9e-06,
+        "output_cost_per_token_above_272k_tokens_priority": 3.6e-05,
         "output_cost_per_token_batches": 6e-06,
         "output_cost_per_token_flex": 6e-06,
         "output_cost_per_token_priority": 2.4e-05,
@@ -29288,16 +29310,19 @@
         "cache_creation_input_token_cost": 2.5e-07,
         "cache_creation_input_token_cost_above_272k_tokens": 5e-07,
         "cache_creation_input_token_cost_above_272k_tokens_flex": 2.5e-07,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 1e-06,
         "cache_creation_input_token_cost_flex": 1.25e-07,
         "cache_creation_input_token_cost_priority": 5e-07,
         "cache_read_input_token_cost": 2e-08,
         "cache_read_input_token_cost_above_272k_tokens": 4e-08,
         "cache_read_input_token_cost_above_272k_tokens_flex": 2e-08,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 8e-08,
         "cache_read_input_token_cost_flex": 1e-08,
         "cache_read_input_token_cost_priority": 4e-08,
         "input_cost_per_token": 2e-07,
         "input_cost_per_token_above_272k_tokens": 4e-07,
         "input_cost_per_token_above_272k_tokens_flex": 2e-07,
+        "input_cost_per_token_above_272k_tokens_priority": 8e-07,
         "input_cost_per_token_batches": 1e-07,
         "input_cost_per_token_flex": 1e-07,
         "input_cost_per_token_priority": 4e-07,
@@ -29309,6 +29334,7 @@
         "output_cost_per_token": 1.2e-06,
         "output_cost_per_token_above_272k_tokens": 1.8e-06,
         "output_cost_per_token_above_272k_tokens_flex": 9e-07,
+        "output_cost_per_token_above_272k_tokens_priority": 3.6e-06,
         "output_cost_per_token_batches": 6e-07,
         "output_cost_per_token_flex": 6e-07,
         "output_cost_per_token_priority": 2.4e-06,
@@ -29548,7 +29574,10 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": true,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": false
+        "supports_minimal_reasoning_effort": false,
+        "input_cost_per_token_above_272k_tokens_flex": 5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 2.25e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 5e-07
     },
     "gpt-5.5-2026-04-23": {
         "cache_read_input_token_cost": 5e-07,
@@ -29602,7 +29631,10 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": true,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": false
+        "supports_minimal_reasoning_effort": false,
+        "input_cost_per_token_above_272k_tokens_flex": 5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 2.25e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 5e-07
     },
     "gpt-5.5-pro": {
         "cache_read_input_token_cost": 3e-06,
@@ -29751,7 +29783,10 @@
         "supports_none_reasoning_effort": true,
         "default_reasoning_effort": "none",
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 2.5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 1.125e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 2.5e-07
     },
     "gpt-5.4-2026-03-05": {
         "cache_read_input_token_cost": 2.5e-07,
@@ -29800,7 +29835,10 @@
         "supports_none_reasoning_effort": true,
         "default_reasoning_effort": "none",
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 2.5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 1.125e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 2.5e-07
     },
     "gpt-5.4-pro": {
         "cache_read_input_token_cost": 3e-06,
@@ -29849,7 +29887,9 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": false,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 3e-05,
+        "output_cost_per_token_above_272k_tokens_flex": 0.000135
     },
     "gpt-5.4-pro-2026-03-05": {
         "cache_read_input_token_cost": 3e-06,
@@ -29898,7 +29938,9 @@
         "supports_web_search": true,
         "supports_none_reasoning_effort": false,
         "supports_xhigh_reasoning_effort": true,
-        "supports_minimal_reasoning_effort": true
+        "supports_minimal_reasoning_effort": true,
+        "input_cost_per_token_above_272k_tokens_flex": 3e-05,
+        "output_cost_per_token_above_272k_tokens_flex": 0.000135
     },
     "gpt-5.4-mini": {
         "cache_read_input_token_cost": 7.5e-08,
@@ -41350,13 +41392,13 @@
         "source": "https://docs.together.ai/docs/serverless-models"
     },
     "together_ai/Qwen/Qwen3.8-2.4T-A95B": {
-        "cache_read_input_token_cost": 5e-07,
-        "input_cost_per_token": 2.5e-06,
+        "cache_read_input_token_cost": 2.5e-07,
+        "input_cost_per_token": 2e-06,
         "litellm_provider": "together_ai",
         "max_input_tokens": 1010000,
         "max_tokens": 1010000,
         "mode": "chat",
-        "output_cost_per_token": 6.25e-06,
+        "output_cost_per_token": 6e-06,
         "source": "https://docs.together.ai/docs/serverless-models",
         "supports_prompt_caching": true
     },
@@ -57558,5 +57600,526 @@
         "supported_endpoints": [
             "/v1/audio/transcriptions"
         ]
+    },
+    "scaleway/glm-5.2": {
+        "input_cost_per_token": 1.8e-06,
+        "litellm_provider": "scaleway",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 16384,
+        "max_tokens": 16384,
+        "mode": "chat",
+        "output_cost_per_token": 5.5e-06,
+        "source": "https://www.scaleway.com/en/pricing/model-as-a-service/",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_vision": false
+    },
+    "scaleway/deepseek-v4-flash-0731": {
+        "cache_read_input_token_cost": 8e-08,
+        "input_cost_per_token": 4e-07,
+        "litellm_provider": "scaleway",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 32768,
+        "max_tokens": 32768,
+        "mode": "chat",
+        "output_cost_per_token": 8e-07,
+        "source": "https://www.scaleway.com/en/pricing/model-as-a-service/",
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_vision": false
+    },
+    "azure_ai/kimi-k2.7-code": {
+        "cache_read_input_token_cost": 1.9e-07,
+        "deprecation_date": "2026-10-03",
+        "input_cost_per_token": 9.5e-07,
+        "litellm_provider": "azure_ai",
+        "max_input_tokens": 262144,
+        "max_output_tokens": 262144,
+        "max_tokens": 262144,
+        "mode": "chat",
+        "output_cost_per_token": 4e-06,
+        "source": "https://azure.microsoft.com/en-us/pricing/details/ai-foundry-models/kimi/",
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
+    "bedrock/us-gov-west-1/nvidia.nemotron-nano-3-30b": {
+        "input_cost_per_token": 7.2e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 262144,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 2.88e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_native_structured_output": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/nvidia.nemotron-nano-12b-v2": {
+        "input_cost_per_token": 2.4e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_system_messages": true,
+        "supports_vision": true
+    },
+    "bedrock/us-gov-west-1/nvidia.nemotron-super-3-120b": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 32768,
+        "max_tokens": 32768,
+        "mode": "chat",
+        "output_cost_per_token": 7.8e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/openai.gpt-oss-20b-1:0": {
+        "input_cost_per_token": 8.4e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3.6e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/openai.gpt-oss-120b-1:0": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-west-1/anthropic.claude-sonnet-5": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 3e-06,
+        "cache_creation_input_token_cost_above_1hr": 4.8e-06,
+        "cache_read_input_token_cost": 2.4e-07,
+        "input_cost_per_token": 2.4e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": false,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock/us-gov-west-1/anthropic.claude-opus-4-8": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 7.5e-06,
+        "cache_creation_input_token_cost_above_1hr": 1.2e-05,
+        "cache_read_input_token_cost": 6e-07,
+        "input_cost_per_token": 6e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": true,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock/us-gov-east-1/nvidia.nemotron-nano-3-30b": {
+        "input_cost_per_token": 7.2e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 262144,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 2.88e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_native_structured_output": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/nvidia.nemotron-nano-12b-v2": {
+        "input_cost_per_token": 2.4e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 8192,
+        "max_tokens": 8192,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_system_messages": true,
+        "supports_vision": true
+    },
+    "bedrock/us-gov-east-1/nvidia.nemotron-super-3-120b": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 256000,
+        "max_output_tokens": 32768,
+        "max_tokens": 32768,
+        "mode": "chat",
+        "output_cost_per_token": 7.8e-07,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/openai.gpt-oss-20b-1:0": {
+        "input_cost_per_token": 8.4e-08,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3.6e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/openai.gpt-oss-120b-1:0": {
+        "input_cost_per_token": 1.8e-07,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 128000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 7.2e-07,
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true
+    },
+    "bedrock/us-gov-east-1/anthropic.claude-sonnet-5": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 3e-06,
+        "cache_creation_input_token_cost_above_1hr": 4.8e-06,
+        "cache_read_input_token_cost": 2.4e-07,
+        "input_cost_per_token": 2.4e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": false,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock/us-gov-east-1/anthropic.claude-opus-4-8": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 7.5e-06,
+        "cache_creation_input_token_cost_above_1hr": 1.2e-05,
+        "cache_read_input_token_cost": 6e-07,
+        "input_cost_per_token": 6e-06,
+        "litellm_provider": "bedrock",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": true,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "bedrock_mantle/us-gov-west-1/openai.gpt-5.6-terra": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 2.64e-06,
+        "input_cost_per_token_above_272k_tokens": 5.28e-06,
+        "cache_creation_input_token_cost": 3.3e-06,
+        "cache_creation_input_token_cost_above_272k_tokens": 6.6e-06,
+        "cache_read_input_token_cost": 2.64e-07,
+        "cache_read_input_token_cost_above_272k_tokens": 5.28e-07,
+        "output_cost_per_token": 1.584e-05,
+        "output_cost_per_token_above_272k_tokens": 2.376e-05
+    },
+    "bedrock_mantle/us-gov-west-1/openai.gpt-5.6-luna": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 2.64e-07,
+        "input_cost_per_token_above_272k_tokens": 5.28e-07,
+        "cache_creation_input_token_cost": 3.3e-07,
+        "cache_creation_input_token_cost_above_272k_tokens": 6.6e-07,
+        "cache_read_input_token_cost": 2.64e-08,
+        "cache_read_input_token_cost_above_272k_tokens": 5.28e-08,
+        "output_cost_per_token": 1.584e-06,
+        "output_cost_per_token_above_272k_tokens": 2.376e-06
+    },
+    "bedrock_mantle/us-gov-west-1/openai.gpt-5.4": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 3.3e-06,
+        "cache_read_input_token_cost": 3.3e-07,
+        "output_cost_per_token": 1.98e-05
+    },
+    "bedrock_mantle/us-gov-west-1/xai.grok-4.3": {
+        "use_openai_responses_path": true,
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 131072,
+        "max_output_tokens": 16384,
+        "max_tokens": 16384,
+        "mode": "chat",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/responses"
+        ],
+        "supports_function_calling": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "source": "https://aws.amazon.com/bedrock/pricing/",
+        "input_cost_per_token": 1.5e-06,
+        "output_cost_per_token": 3e-06,
+        "cache_read_input_token_cost": 2.4e-07
+    },
+    "bedrock_mantle/us-gov-east-1/openai.gpt-5.4": {
+        "litellm_provider": "bedrock_mantle",
+        "max_input_tokens": 1050000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "responses",
+        "use_openai_responses_path": true,
+        "supported_endpoints": [
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_function_calling": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "input_cost_per_token": 3.3e-06,
+        "cache_read_input_token_cost": 3.3e-07,
+        "output_cost_per_token": 1.98e-05
+    },
+    "azure/us-gov/gpt-5.1": {
+        "cache_read_input_token_cost": 1.71875e-07,
+        "default_reasoning_effort": "none",
+        "deprecation_date": "2027-05-15",
+        "input_cost_per_token": 1.71875e-06,
+        "litellm_provider": "azure",
+        "max_input_tokens": 272000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.375e-05,
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/batch",
+            "/v1/responses"
+        ],
+        "supported_modalities": [
+            "text",
+            "image"
+        ],
+        "supported_output_modalities": [
+            "text",
+            "image"
+        ],
+        "supports_function_calling": true,
+        "supports_native_streaming": true,
+        "supports_none_reasoning_effort": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_vision": true
+    },
+    "azure/us-gov/o3-mini": {
+        "cache_read_input_token_cost": 7.57e-07,
+        "deprecation_date": "2026-10-01",
+        "input_cost_per_token": 1.513e-06,
+        "litellm_provider": "azure",
+        "max_input_tokens": 200000,
+        "max_output_tokens": 100000,
+        "max_tokens": 100000,
+        "mode": "chat",
+        "output_cost_per_token": 6.05e-06,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_tool_choice": true,
+        "supports_vision": false
+    },
+    "azure/us-gov/text-embedding-3-large": {
+        "deprecation_date": "2028-02-09",
+        "input_cost_per_token": 1.63e-07,
+        "litellm_provider": "azure",
+        "max_input_tokens": 8191,
+        "max_tokens": 8191,
+        "mode": "embedding",
+        "output_cost_per_token": 0.0
+    },
+    "azure/us-gov/text-embedding-3-small": {
+        "deprecation_date": "2028-02-09",
+        "input_cost_per_token": 2.5e-08,
+        "litellm_provider": "azure",
+        "max_input_tokens": 8191,
+        "max_tokens": 8191,
+        "mode": "embedding",
+        "output_cost_per_token": 0.0
     }
 }
diff --git a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
index 0e1c832ebf5..5c8de19a7e9 100644
--- a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
+++ b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
@@ -1522,7 +1522,7 @@ def test_gpt_5_6_alias_prices_match_sol(local_model_cost_map):
     sol = litellm.model_cost["gpt-5.6-sol"]
 
     cost_fields = sorted(field for field in sol if "cost" in field)
-    assert len(cost_fields) == 23
+    assert len(cost_fields) == 27
 
     for field in cost_fields:
         assert alias.get(field) == sol.get(field), field
@@ -4039,8 +4039,8 @@ def test_fast_service_tier_matches_priority_above_the_context_threshold(_local_m
     )
 
     assert fast == priority
-    assert fast[0] == pytest.approx(300_000 * 8e-06, rel=1e-9)
-    assert fast[1] == pytest.approx(1_000 * 3e-05, rel=1e-9)
+    assert fast[0] == pytest.approx(300_000 * 1.6e-05, rel=1e-9)
+    assert fast[1] == pytest.approx(1_000 * 6e-05, rel=1e-9)
 
 
 def test_priority_reasoning_tokens_bill_at_the_priority_output_rate(_local_model_cost_map):
diff --git a/tests/test_litellm/test_openai_service_tier_long_context_pricing.py b/tests/test_litellm/test_openai_service_tier_long_context_pricing.py
new file mode 100644
index 00000000000..c0860a5b55f
--- /dev/null
+++ b/tests/test_litellm/test_openai_service_tier_long_context_pricing.py
@@ -0,0 +1,156 @@
+import json
+from functools import lru_cache
+from pathlib import Path
+
+import pytest
+
+import litellm
+
+REPO_ROOT = Path(__file__).parents[2]
+MAIN_PATH = REPO_ROOT / "model_prices_and_context_window.json"
+BACKUP_PATH = REPO_ROOT / "litellm" / "model_prices_and_context_window_backup.json"
+
+FLEX_LONG_CONTEXT = {
+    "gpt-5.4": {
+        "input_cost_per_token_above_272k_tokens_flex": 2.5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 1.125e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 2.5e-07,
+    },
+    "gpt-5.4-pro": {
+        "input_cost_per_token_above_272k_tokens_flex": 3e-05,
+        "output_cost_per_token_above_272k_tokens_flex": 0.000135,
+    },
+    "gpt-5.5": {
+        "input_cost_per_token_above_272k_tokens_flex": 5e-06,
+        "output_cost_per_token_above_272k_tokens_flex": 2.25e-05,
+        "cache_read_input_token_cost_above_272k_tokens_flex": 5e-07,
+    },
+}
+
+PRIORITY_LONG_CONTEXT = {
+    "gpt-5.6": {
+        "input_cost_per_token_above_272k_tokens_priority": 1.6e-05,
+        "output_cost_per_token_above_272k_tokens_priority": 6e-05,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 1.6e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 2e-05,
+    },
+    "gpt-5.6-sol": {
+        "input_cost_per_token_above_272k_tokens_priority": 1.6e-05,
+        "output_cost_per_token_above_272k_tokens_priority": 6e-05,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 1.6e-06,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 2e-05,
+    },
+    "gpt-5.6-terra": {
+        "input_cost_per_token_above_272k_tokens_priority": 8e-06,
+        "output_cost_per_token_above_272k_tokens_priority": 3.6e-05,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 8e-07,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 1e-05,
+    },
+    "gpt-5.6-luna": {
+        "input_cost_per_token_above_272k_tokens_priority": 8e-07,
+        "output_cost_per_token_above_272k_tokens_priority": 3.6e-06,
+        "cache_read_input_token_cost_above_272k_tokens_priority": 8e-08,
+        "cache_creation_input_token_cost_above_272k_tokens_priority": 1e-06,
+    },
+}
+
+EXPECTED = {**FLEX_LONG_CONTEXT, **PRIORITY_LONG_CONTEXT}
+
+NO_PUBLISHED_PRIORITY_LONG_CONTEXT = ("gpt-5.4", "gpt-5.5")
+
+
+@pytest.fixture(autouse=True)
+def _local_model_cost_map(monkeypatch: pytest.MonkeyPatch) -> None:
+    monkeypatch.setenv("LITELLM_LOCAL_MODEL_COST_MAP", "True")
+    monkeypatch.setattr(litellm, "model_cost", litellm.get_model_cost_map(url=""))
+
+
+@lru_cache(maxsize=2)
+def _load(path: Path) -> dict[str, dict[str, object]]:
+    with open(path) as f:
+        return json.load(f)
+
+
+@pytest.mark.parametrize("path", [MAIN_PATH, BACKUP_PATH], ids=["main", "backup"])
+@pytest.mark.parametrize("model", sorted(EXPECTED))
+def test_service_tier_long_context_rates_are_published(model: str, path: Path) -> None:
+    """Each tier must carry its own above-272K rates, in both price files."""
+    info = _load(path).get(model)
+    assert info is not None, f"{model} not found in {path.name}"
+    for key, expected in EXPECTED[model].items():
+        assert info.get(key) == pytest.approx(expected), f"{model}.{key} is {info.get(key)!r}, expected {expected!r}"
+
+
+@pytest.mark.parametrize("model", sorted(EXPECTED))
+def test_tier_long_context_rate_is_half_or_double_the_standard(model: str) -> None:
+    """Flex is half the standard long-context rate; priority is double it."""
+    info = _load(MAIN_PATH)[model]
+    tier = "flex" if model in FLEX_LONG_CONTEXT else "priority"
+    ratio = 0.5 if tier == "flex" else 2.0
+    for base in ("input_cost_per_token", "output_cost_per_token"):
+        standard = info[f"{base}_above_272k_tokens"]
+        tiered = info[f"{base}_above_272k_tokens_{tier}"]
+        assert tiered == pytest.approx(standard * ratio), (
+            f"{model}.{base}_above_272k_tokens_{tier} is {tiered!r}, "
+            f"expected {ratio}x the standard long-context rate {standard!r}"
+        )
+
+
+@pytest.mark.parametrize("model", NO_PUBLISHED_PRIORITY_LONG_CONTEXT)
+def test_no_priority_long_context_rates_where_openai_publishes_none(model: str) -> None:
+    """Guard against back-filling a rate OpenAI does not publish."""
+    info = _load(MAIN_PATH)[model]
+    assert "input_cost_per_token_above_272k_tokens_priority" not in info
+
+
+LONG_CONTEXT_PROMPT_TOKENS = 300_000
+COMPLETION_TOKENS = 1_000
+
+TIERED_COST_CASES = [
+    ("gpt-5.4", "flex", 2.5e-06, 1.125e-05),
+    ("gpt-5.4-pro", "flex", 3e-05, 0.000135),
+    ("gpt-5.5", "flex", 5e-06, 2.25e-05),
+    ("gpt-5.6", "priority", 1.6e-05, 6e-05),
+    ("gpt-5.6-sol", "priority", 1.6e-05, 6e-05),
+    ("gpt-5.6-terra", "priority", 8e-06, 3.6e-05),
+    ("gpt-5.6-luna", "priority", 8e-07, 3.6e-06),
+]
+
+
+@pytest.mark.parametrize("model,tier,input_rate,output_rate", TIERED_COST_CASES)
+def test_cost_per_token_bills_long_context_at_the_tier_rate(
+    model: str, tier: str, input_rate: float, output_rate: float
+) -> None:
+    """A prompt over 272K on flex or priority must bill at that tier's long-context rate."""
+    input_cost, output_cost = litellm.cost_per_token(
+        model=model,
+        prompt_tokens=LONG_CONTEXT_PROMPT_TOKENS,
+        completion_tokens=COMPLETION_TOKENS,
+        service_tier=tier,
+    )
+    assert input_cost == pytest.approx(LONG_CONTEXT_PROMPT_TOKENS * input_rate)
+    assert output_cost == pytest.approx(COMPLETION_TOKENS * output_rate)
+
+
+@pytest.mark.parametrize("model,tier,input_rate,output_rate", TIERED_COST_CASES)
+def test_cost_per_token_tier_differs_from_the_standard_long_context_cost(
+    model: str, tier: str, input_rate: float, output_rate: float
+) -> None:
+    """Flex halves the standard long-context bill and priority doubles it."""
+    ratio = 0.5 if tier == "flex" else 2.0
+    standard = sum(
+        litellm.cost_per_token(
+            model=model,
+            prompt_tokens=LONG_CONTEXT_PROMPT_TOKENS,
+            completion_tokens=COMPLETION_TOKENS,
+        )
+    )
+    tiered = sum(
+        litellm.cost_per_token(
+            model=model,
+            prompt_tokens=LONG_CONTEXT_PROMPT_TOKENS,
+            completion_tokens=COMPLETION_TOKENS,
+            service_tier=tier,
+        )
+    )
+    assert tiered == pytest.approx(standard * ratio)
diff --git a/whitelisted_bedrock_models.txt b/whitelisted_bedrock_models.txt
index 7e20081988d..8753d7c3c77 100644
--- a/whitelisted_bedrock_models.txt
+++ b/whitelisted_bedrock_models.txt
@@ -217,3 +217,17 @@ bedrock/us-east-1/zai.glm-5
 bedrock/us-west-2/zai.glm-5
 bedrock/us-gov-east-1/anthropic.claude-haiku-4-5-20251001-v1:0
 bedrock/us-gov-west-1/anthropic.claude-haiku-4-5-20251001-v1:0
+bedrock/us-gov-west-1/nvidia.nemotron-nano-3-30b
+bedrock/us-gov-west-1/nvidia.nemotron-nano-12b-v2
+bedrock/us-gov-west-1/nvidia.nemotron-super-3-120b
+bedrock/us-gov-west-1/openai.gpt-oss-20b-1:0
+bedrock/us-gov-west-1/openai.gpt-oss-120b-1:0
+bedrock/us-gov-west-1/anthropic.claude-sonnet-5
+bedrock/us-gov-west-1/anthropic.claude-opus-4-8
+bedrock/us-gov-east-1/nvidia.nemotron-nano-3-30b
+bedrock/us-gov-east-1/nvidia.nemotron-nano-12b-v2
+bedrock/us-gov-east-1/nvidia.nemotron-super-3-120b
+bedrock/us-gov-east-1/openai.gpt-oss-20b-1:0
+bedrock/us-gov-east-1/openai.gpt-oss-120b-1:0
+bedrock/us-gov-east-1/anthropic.claude-sonnet-5
+bedrock/us-gov-east-1/anthropic.claude-opus-4-8

From 6b83b16559e5ceb4904121bcb90623a5f9f7115c Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Wed, 2 Sep 2026 08:04:14 -0700
Subject: [PATCH 10/16] feat(gemini): day-0 pricing for gemini-3.8-flash

Gemini 3.8 Flash launches today with the same promotional pricing, limits,
and thinking settings as Gemini 3.7 Flash, so the gemini/, vertex_ai/, and
bare cost map entries mirror the 3.7 Flash ones. Regression tests lock the
launch prices, the 4096-token cache minimum, and the gemini-3 thought
signature gate in for the new model.
---
 ...odel_prices_and_context_window_backup.json | 173 ++++++++++++++++++
 model_prices_and_context_window.json          | 173 ++++++++++++++++++
 .../llm_cost_calc/test_llm_cost_calc_utils.py |  45 +++++
 .../test_vertex_ai_gemini_transformation.py   |   3 +
 tests/test_litellm/test_utils.py              |   1 +
 5 files changed, 395 insertions(+)

diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index a3cfb300ea6..cc828e126ad 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -23514,6 +23514,63 @@
         "web_search_billing_unit": "per_query",
         "google_maps_grounding_cost_per_query": 0.014
     },
+    "vertex_ai/gemini-3.8-flash": {
+        "prompt_cache_min_tokens": 4096,
+        "cache_read_input_token_cost": 7.5e-08,
+        "cache_read_input_token_cost_flex": 3.75e-08,
+        "input_cost_per_token": 7.5e-07,
+        "input_cost_per_token_batches": 3.75e-07,
+        "input_cost_per_token_flex": 3.75e-07,
+        "litellm_provider": "vertex_ai",
+        "max_input_tokens": 1048576,
+        "max_output_tokens": 65536,
+        "max_tokens": 65536,
+        "mode": "chat",
+        "output_cost_per_reasoning_token": 3.75e-06,
+        "output_cost_per_token": 3.75e-06,
+        "output_cost_per_token_batches": 1.875e-06,
+        "output_cost_per_token_flex": 1.875e-06,
+        "regional_endpoint_uplift_multiplier": 1.1,
+        "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/completions",
+            "/v1/batch"
+        ],
+        "supported_modalities": [
+            "text",
+            "image",
+            "audio",
+            "video"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_audio_input": true,
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_url_context": true,
+        "supports_video_input": true,
+        "supports_vision": true,
+        "supports_web_search": true,
+        "supports_native_streaming": true,
+        "input_cost_per_token_priority": 1.35e-06,
+        "output_cost_per_token_priority": 6.75e-06,
+        "cache_read_input_token_cost_priority": 1.35e-07,
+        "search_context_cost_per_query": {
+            "search_context_size_low": 0.014,
+            "search_context_size_medium": 0.014,
+            "search_context_size_high": 0.014
+        },
+        "web_search_billing_unit": "per_query",
+        "google_maps_grounding_cost_per_query": 0.014
+    },
     "vertex_ai/gemini-3.1-pro-preview": {
         "prompt_cache_min_tokens": 4096,
         "cache_read_input_token_cost": 2e-07,
@@ -25351,6 +25408,65 @@
         "web_search_billing_unit": "per_query",
         "google_maps_grounding_cost_per_query": 0.014
     },
+    "gemini/gemini-3.8-flash": {
+        "prompt_cache_min_tokens": 4096,
+        "cache_read_input_token_cost": 7.5e-08,
+        "cache_read_input_token_cost_flex": 3.75e-08,
+        "input_cost_per_token": 7.5e-07,
+        "input_cost_per_token_batches": 3.75e-07,
+        "input_cost_per_token_flex": 3.75e-07,
+        "litellm_provider": "gemini",
+        "max_input_tokens": 1048576,
+        "max_output_tokens": 65536,
+        "max_tokens": 65536,
+        "mode": "chat",
+        "output_cost_per_reasoning_token": 3.75e-06,
+        "output_cost_per_token": 3.75e-06,
+        "output_cost_per_token_batches": 1.875e-06,
+        "output_cost_per_token_flex": 1.875e-06,
+        "rpm": 2000,
+        "source": "https://ai.google.dev/gemini-api/docs/pricing",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/completions",
+            "/v1/batch"
+        ],
+        "supported_modalities": [
+            "text",
+            "image",
+            "audio",
+            "video"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_audio_output": false,
+        "supports_audio_input": true,
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_url_context": true,
+        "supports_video_input": true,
+        "supports_vision": true,
+        "supports_web_search": true,
+        "supports_native_streaming": true,
+        "tpm": 800000,
+        "input_cost_per_token_priority": 1.35e-06,
+        "output_cost_per_token_priority": 6.75e-06,
+        "cache_read_input_token_cost_priority": 1.35e-07,
+        "search_context_cost_per_query": {
+            "search_context_size_low": 0.014,
+            "search_context_size_medium": 0.014,
+            "search_context_size_high": 0.014
+        },
+        "web_search_billing_unit": "per_query",
+        "google_maps_grounding_cost_per_query": 0.014
+    },
     "gemini/gemini-omni-flash-preview": {
         "input_cost_per_audio_token": 1.5e-06,
         "input_cost_per_token": 1.5e-06,
@@ -25759,6 +25875,63 @@
         "web_search_billing_unit": "per_query",
         "google_maps_grounding_cost_per_query": 0.014
     },
+    "gemini-3.8-flash": {
+        "prompt_cache_min_tokens": 4096,
+        "cache_read_input_token_cost": 7.5e-08,
+        "cache_read_input_token_cost_flex": 3.75e-08,
+        "input_cost_per_token": 7.5e-07,
+        "input_cost_per_token_batches": 3.75e-07,
+        "input_cost_per_token_flex": 3.75e-07,
+        "litellm_provider": "vertex_ai-language-models",
+        "max_input_tokens": 1048576,
+        "max_output_tokens": 65536,
+        "max_tokens": 65536,
+        "mode": "chat",
+        "output_cost_per_reasoning_token": 3.75e-06,
+        "output_cost_per_token": 3.75e-06,
+        "output_cost_per_token_batches": 1.875e-06,
+        "output_cost_per_token_flex": 1.875e-06,
+        "source": "https://ai.google.dev/gemini-api/docs/pricing",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/completions",
+            "/v1/batch"
+        ],
+        "supported_modalities": [
+            "text",
+            "image",
+            "audio",
+            "video"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_audio_output": false,
+        "supports_audio_input": true,
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_url_context": true,
+        "supports_video_input": true,
+        "supports_vision": true,
+        "supports_web_search": true,
+        "supports_native_streaming": true,
+        "input_cost_per_token_priority": 1.35e-06,
+        "output_cost_per_token_priority": 6.75e-06,
+        "cache_read_input_token_cost_priority": 1.35e-07,
+        "search_context_cost_per_query": {
+            "search_context_size_low": 0.014,
+            "search_context_size_medium": 0.014,
+            "search_context_size_high": 0.014
+        },
+        "web_search_billing_unit": "per_query",
+        "google_maps_grounding_cost_per_query": 0.014
+    },
     "gemini/gemini-2.5-pro-preview-tts": {
         "cache_read_input_token_cost": 1.25e-07,
         "input_cost_per_audio_token": 7e-07,
diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json
index a3cfb300ea6..cc828e126ad 100644
--- a/model_prices_and_context_window.json
+++ b/model_prices_and_context_window.json
@@ -23514,6 +23514,63 @@
         "web_search_billing_unit": "per_query",
         "google_maps_grounding_cost_per_query": 0.014
     },
+    "vertex_ai/gemini-3.8-flash": {
+        "prompt_cache_min_tokens": 4096,
+        "cache_read_input_token_cost": 7.5e-08,
+        "cache_read_input_token_cost_flex": 3.75e-08,
+        "input_cost_per_token": 7.5e-07,
+        "input_cost_per_token_batches": 3.75e-07,
+        "input_cost_per_token_flex": 3.75e-07,
+        "litellm_provider": "vertex_ai",
+        "max_input_tokens": 1048576,
+        "max_output_tokens": 65536,
+        "max_tokens": 65536,
+        "mode": "chat",
+        "output_cost_per_reasoning_token": 3.75e-06,
+        "output_cost_per_token": 3.75e-06,
+        "output_cost_per_token_batches": 1.875e-06,
+        "output_cost_per_token_flex": 1.875e-06,
+        "regional_endpoint_uplift_multiplier": 1.1,
+        "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/completions",
+            "/v1/batch"
+        ],
+        "supported_modalities": [
+            "text",
+            "image",
+            "audio",
+            "video"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_audio_input": true,
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_url_context": true,
+        "supports_video_input": true,
+        "supports_vision": true,
+        "supports_web_search": true,
+        "supports_native_streaming": true,
+        "input_cost_per_token_priority": 1.35e-06,
+        "output_cost_per_token_priority": 6.75e-06,
+        "cache_read_input_token_cost_priority": 1.35e-07,
+        "search_context_cost_per_query": {
+            "search_context_size_low": 0.014,
+            "search_context_size_medium": 0.014,
+            "search_context_size_high": 0.014
+        },
+        "web_search_billing_unit": "per_query",
+        "google_maps_grounding_cost_per_query": 0.014
+    },
     "vertex_ai/gemini-3.1-pro-preview": {
         "prompt_cache_min_tokens": 4096,
         "cache_read_input_token_cost": 2e-07,
@@ -25351,6 +25408,65 @@
         "web_search_billing_unit": "per_query",
         "google_maps_grounding_cost_per_query": 0.014
     },
+    "gemini/gemini-3.8-flash": {
+        "prompt_cache_min_tokens": 4096,
+        "cache_read_input_token_cost": 7.5e-08,
+        "cache_read_input_token_cost_flex": 3.75e-08,
+        "input_cost_per_token": 7.5e-07,
+        "input_cost_per_token_batches": 3.75e-07,
+        "input_cost_per_token_flex": 3.75e-07,
+        "litellm_provider": "gemini",
+        "max_input_tokens": 1048576,
+        "max_output_tokens": 65536,
+        "max_tokens": 65536,
+        "mode": "chat",
+        "output_cost_per_reasoning_token": 3.75e-06,
+        "output_cost_per_token": 3.75e-06,
+        "output_cost_per_token_batches": 1.875e-06,
+        "output_cost_per_token_flex": 1.875e-06,
+        "rpm": 2000,
+        "source": "https://ai.google.dev/gemini-api/docs/pricing",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/completions",
+            "/v1/batch"
+        ],
+        "supported_modalities": [
+            "text",
+            "image",
+            "audio",
+            "video"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_audio_output": false,
+        "supports_audio_input": true,
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_url_context": true,
+        "supports_video_input": true,
+        "supports_vision": true,
+        "supports_web_search": true,
+        "supports_native_streaming": true,
+        "tpm": 800000,
+        "input_cost_per_token_priority": 1.35e-06,
+        "output_cost_per_token_priority": 6.75e-06,
+        "cache_read_input_token_cost_priority": 1.35e-07,
+        "search_context_cost_per_query": {
+            "search_context_size_low": 0.014,
+            "search_context_size_medium": 0.014,
+            "search_context_size_high": 0.014
+        },
+        "web_search_billing_unit": "per_query",
+        "google_maps_grounding_cost_per_query": 0.014
+    },
     "gemini/gemini-omni-flash-preview": {
         "input_cost_per_audio_token": 1.5e-06,
         "input_cost_per_token": 1.5e-06,
@@ -25759,6 +25875,63 @@
         "web_search_billing_unit": "per_query",
         "google_maps_grounding_cost_per_query": 0.014
     },
+    "gemini-3.8-flash": {
+        "prompt_cache_min_tokens": 4096,
+        "cache_read_input_token_cost": 7.5e-08,
+        "cache_read_input_token_cost_flex": 3.75e-08,
+        "input_cost_per_token": 7.5e-07,
+        "input_cost_per_token_batches": 3.75e-07,
+        "input_cost_per_token_flex": 3.75e-07,
+        "litellm_provider": "vertex_ai-language-models",
+        "max_input_tokens": 1048576,
+        "max_output_tokens": 65536,
+        "max_tokens": 65536,
+        "mode": "chat",
+        "output_cost_per_reasoning_token": 3.75e-06,
+        "output_cost_per_token": 3.75e-06,
+        "output_cost_per_token_batches": 1.875e-06,
+        "output_cost_per_token_flex": 1.875e-06,
+        "source": "https://ai.google.dev/gemini-api/docs/pricing",
+        "supported_endpoints": [
+            "/v1/chat/completions",
+            "/v1/completions",
+            "/v1/batch"
+        ],
+        "supported_modalities": [
+            "text",
+            "image",
+            "audio",
+            "video"
+        ],
+        "supported_output_modalities": [
+            "text"
+        ],
+        "supports_audio_output": false,
+        "supports_audio_input": true,
+        "supports_function_calling": true,
+        "supports_parallel_function_calling": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_system_messages": true,
+        "supports_tool_choice": true,
+        "supports_url_context": true,
+        "supports_video_input": true,
+        "supports_vision": true,
+        "supports_web_search": true,
+        "supports_native_streaming": true,
+        "input_cost_per_token_priority": 1.35e-06,
+        "output_cost_per_token_priority": 6.75e-06,
+        "cache_read_input_token_cost_priority": 1.35e-07,
+        "search_context_cost_per_query": {
+            "search_context_size_low": 0.014,
+            "search_context_size_medium": 0.014,
+            "search_context_size_high": 0.014
+        },
+        "web_search_billing_unit": "per_query",
+        "google_maps_grounding_cost_per_query": 0.014
+    },
     "gemini/gemini-2.5-pro-preview-tts": {
         "cache_read_input_token_cost": 1.25e-07,
         "input_cost_per_audio_token": 7e-07,
diff --git a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
index 0e1c832ebf5..0ccb05c67a3 100644
--- a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
+++ b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
@@ -4200,6 +4200,51 @@ def test_generic_cost_per_token_gemini_37_flash(_local_model_cost_map):
     assert completion_cost == pytest.approx(0.001875)
 
 
+GEMINI_38_FLASH_LAUNCH_PRICING = [
+    ("gemini-3.8-flash", 7.5e-07, 3.75e-06, 7.5e-08),
+    ("gemini/gemini-3.8-flash", 7.5e-07, 3.75e-06, 7.5e-08),
+    ("vertex_ai/gemini-3.8-flash", 7.5e-07, 3.75e-06, 7.5e-08),
+]
+
+
+@pytest.mark.parametrize("model,input_cost,output_cost,cache_read_cost", GEMINI_38_FLASH_LAUNCH_PRICING)
+def test_gemini_38_flash_launch_pricing(model, input_cost, output_cost, cache_read_cost, _local_model_cost_map):
+    model_cost_map = litellm.model_cost[model]
+    assert model_cost_map["input_cost_per_token"] == input_cost
+    assert model_cost_map["output_cost_per_token"] == output_cost
+    assert model_cost_map["output_cost_per_reasoning_token"] == output_cost
+    assert model_cost_map["cache_read_input_token_cost"] == cache_read_cost
+    assert model_cost_map["mode"] == "chat"
+    assert model_cost_map["supports_reasoning"] is True
+    assert model_cost_map["supports_function_calling"] is True
+    assert model_cost_map["max_input_tokens"] == 1048576
+
+
+def test_gemini_38_flash_matches_37_flash_promotional_pricing(_local_model_cost_map):
+    for prefix in ("", "gemini/", "vertex_ai/"):
+        assert litellm.model_cost[f"{prefix}gemini-3.8-flash"] == litellm.model_cost[f"{prefix}gemini-3.7-flash"]
+
+
+def test_generic_cost_per_token_gemini_38_flash(_local_model_cost_map):
+    usage = Usage(
+        prompt_tokens=1000,
+        completion_tokens=500,
+        total_tokens=1500,
+        completion_tokens_details=CompletionTokensDetailsWrapper(
+            reasoning_tokens=200,
+            text_tokens=300,
+        ),
+        prompt_tokens_details=PromptTokensDetailsWrapper(text_tokens=1000),
+    )
+    prompt_cost, completion_cost = generic_cost_per_token(
+        model="gemini-3.8-flash",
+        usage=usage,
+        custom_llm_provider="gemini",
+    )
+    assert prompt_cost == pytest.approx(0.00075)
+    assert completion_cost == pytest.approx(0.001875)
+
+
 def test_grok_46_launch_pricing(_local_model_cost_map):
     model_cost_map = litellm.model_cost["xai/grok-4.6"]
     assert model_cost_map["input_cost_per_token"] == 2e-06
diff --git a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_ai_gemini_transformation.py b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_ai_gemini_transformation.py
index 8c1de12e7d9..4679b978f78 100644
--- a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_ai_gemini_transformation.py
+++ b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_ai_gemini_transformation.py
@@ -1096,10 +1096,13 @@ def test_natively_signed_parallel_turn_never_carries_a_placeholder(model):
         "gemini-3.5-flash",
         "gemini-3.6-flash",
         "gemini-3.7-flash",
+        "gemini-3.8-flash",
         "vertex_ai/gemini-3.5-flash",
         "vertex_ai/gemini-3.7-flash",
+        "vertex_ai/gemini-3.8-flash",
         "gemini/gemini-3.5-flash",
         "gemini/gemini-3.7-flash",
+        "gemini/gemini-3.8-flash",
     ],
 )
 def test_placeholder_scoped_to_first_call_across_gemini_3_variants(model):
diff --git a/tests/test_litellm/test_utils.py b/tests/test_litellm/test_utils.py
index 521e91daded..0790b41c349 100644
--- a/tests/test_litellm/test_utils.py
+++ b/tests/test_litellm/test_utils.py
@@ -4655,6 +4655,7 @@ GEMINI_4096_CACHE_MIN_MODELS: Final = tuple(
         "gemini-3.5-flash",
         "gemini-3.6-flash",
         "gemini-3.7-flash",
+        "gemini-3.8-flash",
         "gemini-3.1-pro-preview",
         "gemini-3.1-pro-preview-customtools",
     )

From b76127774059d577229bbc9f74b3bf1b9fef812c Mon Sep 17 00:00:00 2001
From: mateo <mateo@berri.ai>
Date: Wed, 2 Sep 2026 15:09:49 +0000
Subject: [PATCH 11/16] fix(models): drop inherited retirement dates from
 azure/us-gov entries pending a Government schedule source

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 litellm/model_prices_and_context_window_backup.json | 4 ----
 model_prices_and_context_window.json                | 4 ----
 2 files changed, 8 deletions(-)

diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index 87d348f4752..63b88c2a7b4 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -58055,7 +58055,6 @@
     "azure/us-gov/gpt-5.1": {
         "cache_read_input_token_cost": 1.71875e-07,
         "default_reasoning_effort": "none",
-        "deprecation_date": "2027-05-15",
         "input_cost_per_token": 1.71875e-06,
         "litellm_provider": "azure",
         "max_input_tokens": 272000,
@@ -58090,7 +58089,6 @@
     },
     "azure/us-gov/o3-mini": {
         "cache_read_input_token_cost": 7.57e-07,
-        "deprecation_date": "2026-10-01",
         "input_cost_per_token": 1.513e-06,
         "litellm_provider": "azure",
         "max_input_tokens": 200000,
@@ -58105,7 +58103,6 @@
         "supports_vision": false
     },
     "azure/us-gov/text-embedding-3-large": {
-        "deprecation_date": "2028-02-09",
         "input_cost_per_token": 1.63e-07,
         "litellm_provider": "azure",
         "max_input_tokens": 8191,
@@ -58114,7 +58111,6 @@
         "output_cost_per_token": 0.0
     },
     "azure/us-gov/text-embedding-3-small": {
-        "deprecation_date": "2028-02-09",
         "input_cost_per_token": 2.5e-08,
         "litellm_provider": "azure",
         "max_input_tokens": 8191,
diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json
index 87d348f4752..63b88c2a7b4 100644
--- a/model_prices_and_context_window.json
+++ b/model_prices_and_context_window.json
@@ -58055,7 +58055,6 @@
     "azure/us-gov/gpt-5.1": {
         "cache_read_input_token_cost": 1.71875e-07,
         "default_reasoning_effort": "none",
-        "deprecation_date": "2027-05-15",
         "input_cost_per_token": 1.71875e-06,
         "litellm_provider": "azure",
         "max_input_tokens": 272000,
@@ -58090,7 +58089,6 @@
     },
     "azure/us-gov/o3-mini": {
         "cache_read_input_token_cost": 7.57e-07,
-        "deprecation_date": "2026-10-01",
         "input_cost_per_token": 1.513e-06,
         "litellm_provider": "azure",
         "max_input_tokens": 200000,
@@ -58105,7 +58103,6 @@
         "supports_vision": false
     },
     "azure/us-gov/text-embedding-3-large": {
-        "deprecation_date": "2028-02-09",
         "input_cost_per_token": 1.63e-07,
         "litellm_provider": "azure",
         "max_input_tokens": 8191,
@@ -58114,7 +58111,6 @@
         "output_cost_per_token": 0.0
     },
     "azure/us-gov/text-embedding-3-small": {
-        "deprecation_date": "2028-02-09",
         "input_cost_per_token": 2.5e-08,
         "litellm_provider": "azure",
         "max_input_tokens": 8191,

From da23e0241dc82649ff56f2e73e6e156c4e098129 Mon Sep 17 00:00:00 2001
From: mateo <mateo@berri.ai>
Date: Wed, 2 Sep 2026 15:28:53 +0000
Subject: [PATCH 12/16] fix(models): add cloudflare whisper transcription
 pricing and pin govcloud pricing tests

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 ...odel_prices_and_context_window_backup.json |  20 ++
 model_prices_and_context_window.json          |  20 ++
 .../test_bedrock_usgov_pricing.py             | 200 +++++++++++++++---
 ...st_cloudflare_workers_ai_model_metadata.py |  16 ++
 4 files changed, 230 insertions(+), 26 deletions(-)

diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index 63b88c2a7b4..30621a17df3 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -58117,5 +58117,25 @@
         "max_tokens": 8191,
         "mode": "embedding",
         "output_cost_per_token": 0.0
+    },
+    "cloudflare/@cf/openai/whisper": {
+        "input_cost_per_second": 7.5e-06,
+        "litellm_provider": "cloudflare",
+        "mode": "audio_transcription",
+        "output_cost_per_second": 0.0,
+        "source": "https://developers.cloudflare.com/workers-ai/models/whisper/",
+        "supported_endpoints": [
+            "/v1/audio/transcriptions"
+        ]
+    },
+    "cloudflare/@cf/openai/whisper-large-v3-turbo": {
+        "input_cost_per_second": 8.5e-06,
+        "litellm_provider": "cloudflare",
+        "mode": "audio_transcription",
+        "output_cost_per_second": 0.0,
+        "source": "https://developers.cloudflare.com/workers-ai/models/whisper-large-v3-turbo/",
+        "supported_endpoints": [
+            "/v1/audio/transcriptions"
+        ]
     }
 }
diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json
index 63b88c2a7b4..30621a17df3 100644
--- a/model_prices_and_context_window.json
+++ b/model_prices_and_context_window.json
@@ -58117,5 +58117,25 @@
         "max_tokens": 8191,
         "mode": "embedding",
         "output_cost_per_token": 0.0
+    },
+    "cloudflare/@cf/openai/whisper": {
+        "input_cost_per_second": 7.5e-06,
+        "litellm_provider": "cloudflare",
+        "mode": "audio_transcription",
+        "output_cost_per_second": 0.0,
+        "source": "https://developers.cloudflare.com/workers-ai/models/whisper/",
+        "supported_endpoints": [
+            "/v1/audio/transcriptions"
+        ]
+    },
+    "cloudflare/@cf/openai/whisper-large-v3-turbo": {
+        "input_cost_per_second": 8.5e-06,
+        "litellm_provider": "cloudflare",
+        "mode": "audio_transcription",
+        "output_cost_per_second": 0.0,
+        "source": "https://developers.cloudflare.com/workers-ai/models/whisper-large-v3-turbo/",
+        "supported_endpoints": [
+            "/v1/audio/transcriptions"
+        ]
     }
 }
diff --git a/tests/test_litellm/test_bedrock_usgov_pricing.py b/tests/test_litellm/test_bedrock_usgov_pricing.py
index 6b3312b5cc4..f9e8fd4c46c 100644
--- a/tests/test_litellm/test_bedrock_usgov_pricing.py
+++ b/tests/test_litellm/test_bedrock_usgov_pricing.py
@@ -26,9 +26,7 @@ import pytest
 
 @pytest.fixture(scope="module")
 def model_data():
-    json_path = os.path.join(
-        os.path.dirname(__file__), "../../model_prices_and_context_window.json"
-    )
+    json_path = os.path.join(os.path.dirname(__file__), "../../model_prices_and_context_window.json")
     with open(json_path) as f:
         return json.load(f)
 
@@ -51,21 +49,14 @@ def test_usgov_sonnet_4_5_pricing(model_data, model_key):
     info = model_data[model_key]
 
     assert info["input_cost_per_token"] == 3.6e-06, (
-        f"{model_key}: input_cost_per_token should be $3.60/MTok "
-        f"(got {info['input_cost_per_token']})"
+        f"{model_key}: input_cost_per_token should be $3.60/MTok (got {info['input_cost_per_token']})"
     )
-    assert (
-        info["output_cost_per_token"] == 1.8e-05
-    ), f"{model_key}: output_cost_per_token should be $18.00/MTok"
-    assert (
-        info["cache_creation_input_token_cost"] == 4.5e-06
-    ), f"{model_key}: 5m cache write should be $4.50/MTok"
-    assert (
-        info["cache_creation_input_token_cost_above_1hr"] == 7.2e-06
-    ), f"{model_key}: 1h cache write should be $7.20/MTok"
-    assert (
-        info["cache_read_input_token_cost"] == 3.6e-07
-    ), f"{model_key}: cache read should be $0.36/MTok"
+    assert info["output_cost_per_token"] == 1.8e-05, f"{model_key}: output_cost_per_token should be $18.00/MTok"
+    assert info["cache_creation_input_token_cost"] == 4.5e-06, f"{model_key}: 5m cache write should be $4.50/MTok"
+    assert info["cache_creation_input_token_cost_above_1hr"] == 7.2e-06, (
+        f"{model_key}: 1h cache write should be $7.20/MTok"
+    )
+    assert info["cache_read_input_token_cost"] == 3.6e-07, f"{model_key}: cache read should be $0.36/MTok"
 
 
 def test_usgov_carries_20_percent_premium_over_global(model_data):
@@ -84,9 +75,7 @@ def test_usgov_carries_20_percent_premium_over_global(model_data):
         "cache_read_input_token_cost",
     ):
         ratio = usgov_info[field] / global_info[field]
-        assert (
-            abs(ratio - 1.2) < 1e-9
-        ), f"{field}: us-gov / global ratio is {ratio}, expected 1.2"
+        assert abs(ratio - 1.2) < 1e-9, f"{field}: us-gov / global ratio is {ratio}, expected 1.2"
 
 
 # The us-gov.anthropic.* cross-region inference profile is the only us-gov
@@ -112,9 +101,7 @@ def test_usgov_cross_region_above_200k_carries_gov_premium(model_data, field, ex
     """
     info = model_data[USGOV_CROSS_REGION_KEY]
     assert field in info, f"{USGOV_CROSS_REGION_KEY}: missing field {field}"
-    assert (
-        info[field] == expected
-    ), f"{USGOV_CROSS_REGION_KEY}: {field} should be {expected} (got {info[field]})"
+    assert info[field] == expected, f"{USGOV_CROSS_REGION_KEY}: {field} should be {expected} (got {info[field]})"
 
 
 def test_usgov_cross_region_above_200k_ratio_to_global(model_data):
@@ -127,6 +114,167 @@ def test_usgov_cross_region_above_200k_ratio_to_global(model_data):
     usgov_info = model_data[USGOV_CROSS_REGION_KEY]
     for field in EXPECTED_USGOV_ABOVE_200K:
         ratio = usgov_info[field] / global_info[field]
-        assert (
-            abs(ratio - 1.2) < 1e-9
-        ), f"{field}: us-gov / global ratio is {ratio}, expected 1.2"
+        assert abs(ratio - 1.2) < 1e-9, f"{field}: us-gov / global ratio is {ratio}, expected 1.2"
+
+
+CLAUDE_GOV_EXPECTED = {
+    "anthropic.claude-sonnet-5": {
+        "input_cost_per_token": 2.4e-06,
+        "output_cost_per_token": 1.2e-05,
+        "cache_creation_input_token_cost": 3e-06,
+        "cache_creation_input_token_cost_above_1hr": 4.8e-06,
+        "cache_read_input_token_cost": 2.4e-07,
+    },
+    "anthropic.claude-opus-4-8": {
+        "input_cost_per_token": 6e-06,
+        "output_cost_per_token": 3e-05,
+        "cache_creation_input_token_cost": 7.5e-06,
+        "cache_creation_input_token_cost_above_1hr": 1.2e-05,
+        "cache_read_input_token_cost": 6e-07,
+    },
+}
+
+
+@pytest.mark.parametrize("base_key", CLAUDE_GOV_EXPECTED)
+@pytest.mark.parametrize("region", ["us-gov-east-1", "us-gov-west-1"])
+def test_usgov_claude_sonnet5_opus48_pricing(model_data, region, base_key):
+    """Sonnet 5 and Opus 4.8 gov entries must match the rates AWS publishes
+    for both GovCloud regions on the Bedrock pricing page (1.2x global).
+    """
+    gov_key = f"bedrock/{region}/{base_key}"
+    assert gov_key in model_data, f"Missing model entry: {gov_key}"
+    info = model_data[gov_key]
+    for field, expected in CLAUDE_GOV_EXPECTED[base_key].items():
+        assert info[field] == expected, f"{gov_key}: {field} should be {expected} (got {info[field]})"
+        ratio = info[field] / model_data[base_key][field]
+        assert abs(ratio - 1.2) < 1e-9, f"{gov_key}: {field} gov/global ratio is {ratio}, expected 1.2"
+
+
+CONVERSE_GOV_EXPECTED = {
+    "nvidia.nemotron-nano-3-30b": (7.2e-08, 2.88e-07),
+    "nvidia.nemotron-nano-12b-v2": (2.4e-07, 7.2e-07),
+    "nvidia.nemotron-super-3-120b": (1.8e-07, 7.8e-07),
+    "openai.gpt-oss-20b-1:0": (8.4e-08, 3.6e-07),
+    "openai.gpt-oss-120b-1:0": (1.8e-07, 7.2e-07),
+}
+
+
+@pytest.mark.parametrize("base_key", CONVERSE_GOV_EXPECTED)
+@pytest.mark.parametrize("region", ["us-gov-east-1", "us-gov-west-1"])
+def test_usgov_converse_model_pricing(model_data, region, base_key):
+    """Nemotron and gpt-oss gov entries must match the AWS Bedrock offer file,
+    which prices both GovCloud regions identically at 1.2x commercial.
+    """
+    gov_key = f"bedrock/{region}/{base_key}"
+    assert gov_key in model_data, f"Missing model entry: {gov_key}"
+    info = model_data[gov_key]
+    expected_input, expected_output = CONVERSE_GOV_EXPECTED[base_key]
+    assert info["input_cost_per_token"] == expected_input
+    assert info["output_cost_per_token"] == expected_output
+    assert info["litellm_provider"] == "bedrock"
+    base = model_data[base_key]
+    assert abs(info["input_cost_per_token"] / base["input_cost_per_token"] - 1.2) < 1e-9
+    assert abs(info["output_cost_per_token"] / base["output_cost_per_token"] - 1.2) < 1e-9
+
+
+def test_usgov_west_llama3_8b_output_price_fixed(model_data):
+    """The us-gov-west-1 llama3-8b entry carried the 70B output rate ($2.65/MTok);
+    the AWS Bedrock offer file prices output at $0.60/MTok. AWS lists the model
+    in us-gov-west-1 only, so there is no east entry to check.
+    """
+    info = model_data["bedrock/us-gov-west-1/meta.llama3-8b-instruct-v1:0"]
+    assert info["input_cost_per_token"] == 3e-07
+    assert info["output_cost_per_token"] == 6e-07
+
+
+MANTLE_GOV_TIERED_EXPECTED = {
+    "openai.gpt-5.6-luna": {
+        "input_cost_per_token": 2.64e-07,
+        "input_cost_per_token_above_272k_tokens": 5.28e-07,
+        "cache_creation_input_token_cost": 3.3e-07,
+        "cache_creation_input_token_cost_above_272k_tokens": 6.6e-07,
+        "cache_read_input_token_cost": 2.64e-08,
+        "cache_read_input_token_cost_above_272k_tokens": 5.28e-08,
+        "output_cost_per_token": 1.584e-06,
+        "output_cost_per_token_above_272k_tokens": 2.376e-06,
+    },
+    "openai.gpt-5.6-terra": {
+        "input_cost_per_token": 2.64e-06,
+        "input_cost_per_token_above_272k_tokens": 5.28e-06,
+        "cache_creation_input_token_cost": 3.3e-06,
+        "cache_creation_input_token_cost_above_272k_tokens": 6.6e-06,
+        "cache_read_input_token_cost": 2.64e-07,
+        "cache_read_input_token_cost_above_272k_tokens": 5.28e-07,
+        "output_cost_per_token": 1.584e-05,
+        "output_cost_per_token_above_272k_tokens": 2.376e-05,
+    },
+}
+
+
+@pytest.mark.parametrize("model", MANTLE_GOV_TIERED_EXPECTED)
+def test_usgov_west_mantle_terra_luna_pricing(model_data, model):
+    """Terra and Luna carry 1.2x commercial across every tier in the
+    us-gov-west-1 offer file; the us-gov-east-1 offer file has no SKUs for them.
+    """
+    gov_key = f"bedrock_mantle/us-gov-west-1/{model}"
+    assert gov_key in model_data, f"Missing model entry: {gov_key}"
+    info = model_data[gov_key]
+    for field, expected in MANTLE_GOV_TIERED_EXPECTED[model].items():
+        assert info[field] == expected, f"{gov_key}: {field} should be {expected} (got {info[field]})"
+    assert info["litellm_provider"] == "bedrock_mantle"
+    assert f"bedrock_mantle/us-gov-east-1/{model}" not in model_data
+
+
+@pytest.mark.parametrize("region", ["us-gov-east-1", "us-gov-west-1"])
+def test_usgov_mantle_gpt_5_4_pricing_has_no_long_context_tier(model_data, region):
+    """gpt-5.4 gov rates come from the offer file, which publishes only the
+    standard tier in GovCloud: no long-context SKUs exist there, unlike commercial.
+    """
+    gov_key = f"bedrock_mantle/{region}/openai.gpt-5.4"
+    assert gov_key in model_data, f"Missing model entry: {gov_key}"
+    info = model_data[gov_key]
+    assert info["input_cost_per_token"] == 3.3e-06
+    assert info["cache_read_input_token_cost"] == 3.3e-07
+    assert info["output_cost_per_token"] == 1.98e-05
+    assert not any(field.endswith("_above_272k_tokens") for field in info)
+
+
+def test_usgov_mantle_grok_4_3_west_only(model_data):
+    """grok-4.3 is priced in the us-gov-west-1 offer file only; the east offer
+    file carries grok-4.6 instead.
+    """
+    info = model_data["bedrock_mantle/us-gov-west-1/xai.grok-4.3"]
+    assert info["input_cost_per_token"] == 1.5e-06
+    assert info["output_cost_per_token"] == 3e-06
+    assert info["cache_read_input_token_cost"] == 2.4e-07
+    assert "bedrock_mantle/us-gov-east-1/xai.grok-4.3" not in model_data
+
+
+AZURE_GOV_EXPECTED = {
+    "azure/us-gov/gpt-5.1": {
+        "input_cost_per_token": 1.71875e-06,
+        "cache_read_input_token_cost": 1.71875e-07,
+        "output_cost_per_token": 1.375e-05,
+    },
+    "azure/us-gov/o3-mini": {
+        "input_cost_per_token": 1.513e-06,
+        "cache_read_input_token_cost": 7.57e-07,
+        "output_cost_per_token": 6.05e-06,
+    },
+    "azure/us-gov/text-embedding-3-large": {"input_cost_per_token": 1.63e-07},
+    "azure/us-gov/text-embedding-3-small": {"input_cost_per_token": 2.5e-08},
+}
+
+
+@pytest.mark.parametrize("gov_key", AZURE_GOV_EXPECTED)
+def test_azure_usgov_pricing(model_data, gov_key):
+    """Azure Government meters from the Azure retail prices API
+    (usgovvirginia/usgovarizona, serviceName 'Foundry Models'). No Government
+    retirement schedule is published, so these entries carry no deprecation_date.
+    """
+    assert gov_key in model_data, f"Missing model entry: {gov_key}"
+    info = model_data[gov_key]
+    for field, expected in AZURE_GOV_EXPECTED[gov_key].items():
+        assert info[field] == expected, f"{gov_key}: {field} should be {expected} (got {info[field]})"
+    assert info["litellm_provider"] == "azure"
+    assert "deprecation_date" not in info
diff --git a/tests/test_litellm/test_cloudflare_workers_ai_model_metadata.py b/tests/test_litellm/test_cloudflare_workers_ai_model_metadata.py
index 9ca4515239a..e33bcfb8378 100644
--- a/tests/test_litellm/test_cloudflare_workers_ai_model_metadata.py
+++ b/tests/test_litellm/test_cloudflare_workers_ai_model_metadata.py
@@ -75,6 +75,22 @@ def test_additional_current_models_are_present():
         assert entry["output_cost_per_token"] > 0
 
 
+@pytest.mark.parametrize(
+    "key, published_price_per_audio_minute",
+    [
+        ("cloudflare/@cf/openai/whisper", 0.00045),
+        ("cloudflare/@cf/openai/whisper-large-v3-turbo", 0.00051),
+    ],
+)
+def test_whisper_transcription_pricing_is_stored_per_second(key, published_price_per_audio_minute):
+    entry = litellm.model_cost[key]
+    assert entry["litellm_provider"] == "cloudflare"
+    assert entry["mode"] == "audio_transcription"
+    assert entry["supported_endpoints"] == ["/v1/audio/transcriptions"]
+    assert entry["output_cost_per_second"] == 0.0
+    assert entry["input_cost_per_second"] == pytest.approx(published_price_per_audio_minute / 60)
+
+
 def test_root_and_backup_have_identical_cloudflare_keys():
     if not os.path.exists(ROOT_MAP):
         pytest.skip("root cost map only ships in source checkouts")

From 69cd1bada6249889a9412155a6086faea65bfe6c Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Wed, 2 Sep 2026 08:46:28 -0700
Subject: [PATCH 13/16] test(gemini): compare gemini-3.8-flash to 3.7 flash
 field by field

---
 .../llm_cost_calc/test_llm_cost_calc_utils.py | 41 +++++++++++++++++--
 1 file changed, 38 insertions(+), 3 deletions(-)

diff --git a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
index 0ccb05c67a3..9b3e60764e3 100644
--- a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
+++ b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
@@ -4220,9 +4220,44 @@ def test_gemini_38_flash_launch_pricing(model, input_cost, output_cost, cache_re
     assert model_cost_map["max_input_tokens"] == 1048576
 
 
-def test_gemini_38_flash_matches_37_flash_promotional_pricing(_local_model_cost_map):
-    for prefix in ("", "gemini/", "vertex_ai/"):
-        assert litellm.model_cost[f"{prefix}gemini-3.8-flash"] == litellm.model_cost[f"{prefix}gemini-3.7-flash"]
+GEMINI_38_FLASH_FIELDS_SHARED_WITH_37_FLASH = (
+    "input_cost_per_token",
+    "output_cost_per_token",
+    "output_cost_per_reasoning_token",
+    "cache_read_input_token_cost",
+    "input_cost_per_token_batches",
+    "output_cost_per_token_batches",
+    "input_cost_per_token_flex",
+    "output_cost_per_token_flex",
+    "cache_read_input_token_cost_flex",
+    "input_cost_per_token_priority",
+    "output_cost_per_token_priority",
+    "cache_read_input_token_cost_priority",
+    "search_context_cost_per_query",
+    "google_maps_grounding_cost_per_query",
+    "prompt_cache_min_tokens",
+    "max_input_tokens",
+    "max_output_tokens",
+    "supports_reasoning",
+    "supports_function_calling",
+    "supports_prompt_caching",
+    "supports_vision",
+    "supports_pdf_input",
+    "supports_audio_input",
+    "supports_video_input",
+    "supports_response_schema",
+    "supports_tool_choice",
+    "supports_web_search",
+    "supports_url_context",
+)
+
+
+@pytest.mark.parametrize("prefix", ["", "gemini/", "vertex_ai/"])
+def test_gemini_38_flash_matches_37_flash_promotional_pricing(prefix, _local_model_cost_map):
+    new_model = litellm.model_cost[f"{prefix}gemini-3.8-flash"]
+    old_model = litellm.model_cost[f"{prefix}gemini-3.7-flash"]
+    for field in GEMINI_38_FLASH_FIELDS_SHARED_WITH_37_FLASH:
+        assert new_model[field] == old_model[field], field
 
 
 def test_generic_cost_per_token_gemini_38_flash(_local_model_cost_map):

From de80e3afe448237c69a535517ca88c12d079ee6d Mon Sep 17 00:00:00 2001
From: Yassin Kortam <yassin@berri.ai>
Date: Wed, 2 Sep 2026 10:19:03 -0700
Subject: [PATCH 14/16] fix(helm): scale the classic chart's HPA out at the
 documented 60 percent CPU (#35975)

* fix(helm): scale the classic chart's HPA out at the documented 60 percent CPU

The litellm-helm chart shipped targetCPUUtilizationPercentage: 80, which is
unexamined helm create scaffold rather than a chosen number. It arrived packaged
with the stock minReplicas: 1, maxReplicas: 100, a commented-out
targetMemoryUtilizationPercentage: 80, and the boilerplate "such as Minikube"
comment, the same provenance as the 128Mi resource example this file just
corrected.

60 is the documented recommendation. The mechanism behind it is scale-up lag:
the chart's own startupProbe is failureThreshold: 30 times periodSeconds: 10, so
a replica can take up to 300 seconds to become ready, and a pod added at 80
percent utilization arrives minutes after saturation.

The memory target stays commented out on purpose. The prisma query engine's
resident memory is a high-water mark that ratchets to the pod's worst-ever write
and is never returned, so a memory-target HPA reads the largest write a pod ever
did rather than what it is doing now, and replicas ratchet up without scaling
back in.

hpa_tests.yaml carried its second suite after a YAML document separator, and
helm-unittest loads only the first document per file, so that suite never ran;
an assertion planted in it still passed. Fold it into the one live suite and add
coverage pinning the rendered CPU target, the absence of a memory metric by
default, and that overrides still take effect.

Bump the chart to 1.1.2, since rendered output changes for anyone running with
autoscaling enabled.

* fix(helm): bump litellm-helm to 1.1.3 after rebase onto 1.1.2

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

---------

Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 helm/litellm-helm/Chart.yaml           |  2 +-
 helm/litellm-helm/tests/hpa_tests.yaml | 42 ++++++++++++++++++++++----
 helm/litellm-helm/values.yaml          | 11 ++++++-
 3 files changed, 47 insertions(+), 8 deletions(-)

diff --git a/helm/litellm-helm/Chart.yaml b/helm/litellm-helm/Chart.yaml
index 3959d85edf3..a3cb388ffc6 100644
--- a/helm/litellm-helm/Chart.yaml
+++ b/helm/litellm-helm/Chart.yaml
@@ -18,7 +18,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 1.1.2
+version: 1.1.3
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to
diff --git a/helm/litellm-helm/tests/hpa_tests.yaml b/helm/litellm-helm/tests/hpa_tests.yaml
index ec18c3591d3..cd062dd5971 100644
--- a/helm/litellm-helm/tests/hpa_tests.yaml
+++ b/helm/litellm-helm/tests/hpa_tests.yaml
@@ -1,4 +1,4 @@
-suite: "hpa with behavior"
+suite: "hpa"
 templates:
   - hpa.yaml
 tests:
@@ -23,14 +23,44 @@ tests:
       - equal: { path: spec.behavior.scaleUp.stabilizationWindowSeconds, value: 60 }
       - equal: { path: spec.behavior.scaleDown.stabilizationWindowSeconds, value: 90 }
 
----
-suite: "hpa without behavior"
-templates:
-  - hpa.yaml
-tests:
   - it: "does not render behavior when not set"
     set:
       autoscaling.enabled: true
     asserts:
       - isKind: { of: HorizontalPodAutoscaler }
       - isNull: { path: spec.behavior }
+
+  - it: "scales on cpu at the documented 60 percent by default"
+    set:
+      autoscaling.enabled: true
+    asserts:
+      - isKind: { of: HorizontalPodAutoscaler }
+      - equal: { path: "spec.metrics[0].resource.name", value: cpu }
+      - equal: { path: "spec.metrics[0].resource.target.type", value: Utilization }
+      - equal: { path: "spec.metrics[0].resource.target.averageUtilization", value: 60 }
+
+  - it: "does not scale on memory by default"
+    set:
+      autoscaling.enabled: true
+    asserts:
+      - lengthEqual: { path: spec.metrics, count: 1 }
+
+  - it: "honours an explicit cpu target override"
+    set:
+      autoscaling.enabled: true
+      autoscaling.targetCPUUtilizationPercentage: 75
+    asserts:
+      - equal: { path: "spec.metrics[0].resource.target.averageUtilization", value: 75 }
+
+  - it: "renders a memory metric only when a memory target is set"
+    set:
+      autoscaling.enabled: true
+      autoscaling.targetMemoryUtilizationPercentage: 80
+    asserts:
+      - lengthEqual: { path: spec.metrics, count: 2 }
+      - equal: { path: "spec.metrics[1].resource.name", value: memory }
+      - equal: { path: "spec.metrics[1].resource.target.averageUtilization", value: 80 }
+
+  - it: "renders no hpa when autoscaling is disabled"
+    asserts:
+      - hasDocuments: { count: 0 }
diff --git a/helm/litellm-helm/values.yaml b/helm/litellm-helm/values.yaml
index f8df98de102..637be2322e3 100644
--- a/helm/litellm-helm/values.yaml
+++ b/helm/litellm-helm/values.yaml
@@ -200,7 +200,16 @@ autoscaling:
   enabled: false
   minReplicas: 1
   maxReplicas: 100
-  targetCPUUtilizationPercentage: 80
+  # 60 is the documented recommendation. See "Recommended Machine Specifications"
+  # in https://docs.litellm.ai/docs/proxy/prod. A new replica clears the startupProbe
+  # above only after up to failureThreshold x periodSeconds = 300 seconds, so a target
+  # high enough to trip near saturation adds capacity minutes after it was needed.
+  targetCPUUtilizationPercentage: 60
+  # Deliberately left unset rather than given a value. The prisma query engine's
+  # resident memory is a high-water mark that ratchets to the pod's worst-ever write
+  # and is never returned, so a memory target reads the largest write a pod ever did
+  # rather than what it is doing now, and replicas ratchet up without scaling back in.
+  # Memory is a floor to provision under 'resources', not a signal to scale on.
   # targetMemoryUtilizationPercentage: 80
   # behavior: {}
 

From 7a35c34303e944f68e69299346ec04371b5f59c7 Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Wed, 2 Sep 2026 10:30:14 -0700
Subject: [PATCH 15/16] fix(models): add the us-gov. geo inference profile keys
 for Claude Sonnet 5 and Opus 4.8

---
 ...odel_prices_and_context_window_backup.json | 64 +++++++++++++++++++
 model_prices_and_context_window.json          | 64 +++++++++++++++++++
 .../test_bedrock_usgov_pricing.py             | 19 ++++--
 3 files changed, 142 insertions(+), 5 deletions(-)

diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index 30621a17df3..28c503966f3 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -42008,6 +42008,70 @@
         "supports_parallel_tool_use_config": true,
         "prompt_cache_min_tokens": 1024
     },
+    "us-gov.anthropic.claude-sonnet-5": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 3e-06,
+        "cache_creation_input_token_cost_above_1hr": 4.8e-06,
+        "cache_read_input_token_cost": 2.4e-07,
+        "input_cost_per_token": 2.4e-06,
+        "litellm_provider": "bedrock_converse",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": false,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "us-gov.anthropic.claude-opus-4-8": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 7.5e-06,
+        "cache_creation_input_token_cost_above_1hr": 1.2e-05,
+        "cache_read_input_token_cost": 6e-07,
+        "input_cost_per_token": 6e-06,
+        "litellm_provider": "bedrock_converse",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": true,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
     "au.anthropic.claude-haiku-4-5-20251001-v1:0": {
         "cache_creation_input_token_cost": 1.375e-06,
         "cache_creation_input_token_cost_above_1hr": 2.2e-06,
diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json
index 30621a17df3..28c503966f3 100644
--- a/model_prices_and_context_window.json
+++ b/model_prices_and_context_window.json
@@ -42008,6 +42008,70 @@
         "supports_parallel_tool_use_config": true,
         "prompt_cache_min_tokens": 1024
     },
+    "us-gov.anthropic.claude-sonnet-5": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 3e-06,
+        "cache_creation_input_token_cost_above_1hr": 4.8e-06,
+        "cache_read_input_token_cost": 2.4e-07,
+        "input_cost_per_token": 2.4e-06,
+        "litellm_provider": "bedrock_converse",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 1.2e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": false,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
+    "us-gov.anthropic.claude-opus-4-8": {
+        "bedrock_converse_supports_strict_tools": false,
+        "bedrock_output_config_effort_ceiling": "xhigh",
+        "cache_creation_input_token_cost": 7.5e-06,
+        "cache_creation_input_token_cost_above_1hr": 1.2e-05,
+        "cache_read_input_token_cost": 6e-07,
+        "input_cost_per_token": 6e-06,
+        "litellm_provider": "bedrock_converse",
+        "max_input_tokens": 1000000,
+        "max_output_tokens": 128000,
+        "max_tokens": 128000,
+        "mode": "chat",
+        "output_cost_per_token": 3e-05,
+        "prompt_cache_min_tokens": 1024,
+        "supports_adaptive_thinking": true,
+        "supports_assistant_prefill": false,
+        "supports_computer_use": true,
+        "supports_function_calling": true,
+        "supports_max_reasoning_effort": true,
+        "supports_mid_conversation_system": true,
+        "supports_native_structured_output": true,
+        "supports_output_config": true,
+        "supports_parallel_tool_use_config": true,
+        "supports_pdf_input": true,
+        "supports_prompt_caching": true,
+        "supports_reasoning": true,
+        "supports_response_schema": true,
+        "supports_sampling_params": false,
+        "supports_tool_choice": true,
+        "supports_vision": true,
+        "supports_xhigh_reasoning_effort": true
+    },
     "au.anthropic.claude-haiku-4-5-20251001-v1:0": {
         "cache_creation_input_token_cost": 1.375e-06,
         "cache_creation_input_token_cost_above_1hr": 2.2e-06,
diff --git a/tests/test_litellm/test_bedrock_usgov_pricing.py b/tests/test_litellm/test_bedrock_usgov_pricing.py
index f9e8fd4c46c..f7d95ecda01 100644
--- a/tests/test_litellm/test_bedrock_usgov_pricing.py
+++ b/tests/test_litellm/test_bedrock_usgov_pricing.py
@@ -135,15 +135,24 @@ CLAUDE_GOV_EXPECTED = {
 }
 
 
+USGOV_CLAUDE_KEY_TEMPLATES = {
+    "bedrock/us-gov-east-1/{base_key}": "bedrock",
+    "bedrock/us-gov-west-1/{base_key}": "bedrock",
+    "us-gov.{base_key}": "bedrock_converse",
+}
+
+
 @pytest.mark.parametrize("base_key", CLAUDE_GOV_EXPECTED)
-@pytest.mark.parametrize("region", ["us-gov-east-1", "us-gov-west-1"])
-def test_usgov_claude_sonnet5_opus48_pricing(model_data, region, base_key):
-    """Sonnet 5 and Opus 4.8 gov entries must match the rates AWS publishes
-    for both GovCloud regions on the Bedrock pricing page (1.2x global).
+@pytest.mark.parametrize("key_template,expected_provider", USGOV_CLAUDE_KEY_TEMPLATES.items())
+def test_usgov_claude_sonnet5_opus48_pricing(model_data, key_template, expected_provider, base_key):
+    """Sonnet 5 and Opus 4.8 gov entries, both in-region keys and the us-gov.
+    geo inference profile the model cards list for GovCloud, must match the
+    rates AWS publishes on the Bedrock pricing page (1.2x global).
     """
-    gov_key = f"bedrock/{region}/{base_key}"
+    gov_key = key_template.format(base_key=base_key)
     assert gov_key in model_data, f"Missing model entry: {gov_key}"
     info = model_data[gov_key]
+    assert info["litellm_provider"] == expected_provider
     for field, expected in CLAUDE_GOV_EXPECTED[base_key].items():
         assert info[field] == expected, f"{gov_key}: {field} should be {expected} (got {info[field]})"
         ratio = info[field] / model_data[base_key][field]

From ffc0a8e428a4d8af7e5b130bddb4f0f9cf0cb229 Mon Sep 17 00:00:00 2001
From: "devin-ai-integration[bot]"
 <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Date: Wed, 2 Sep 2026 10:51:17 -0700
Subject: [PATCH 16/16] fix: run access group key sync UPDATEs on the writer,
 not the read replica (#39128)

Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 .../access_group_key_sync.py                  |  6 +-
 .../test_access_group_key_sync.py             | 57 +++++++++++++++++++
 2 files changed, 61 insertions(+), 2 deletions(-)
 create mode 100644 tests/test_litellm/proxy/management_helpers/test_access_group_key_sync.py

diff --git a/litellm/proxy/management_helpers/access_group_key_sync.py b/litellm/proxy/management_helpers/access_group_key_sync.py
index 5d43cb29978..c9f93fae0d9 100644
--- a/litellm/proxy/management_helpers/access_group_key_sync.py
+++ b/litellm/proxy/management_helpers/access_group_key_sync.py
@@ -38,6 +38,7 @@ from litellm.proxy._types import (
 from litellm.proxy.auth.auth_checks import (
     _delete_cache_access_object,  # pyright: ignore[reportPrivateUsage]  # the access-group endpoints reach for this same cache primitive
 )
+from litellm.proxy.db.routing_prisma_wrapper import WriterPinnedClient
 from litellm.repositories.table_repositories import AccessGroupRepository
 
 
@@ -72,8 +73,9 @@ _REPOINT_KEY_SQL: Final = (
 
 
 def _raw_executor(prisma_client: object) -> _RawExecutor:
-    """Narrow the untyped Prisma client down to the raw-query call this module makes."""
-    return AccessGroupRepository(prisma_client).prisma_client.db  # pyright: ignore[reportAny]  # untyped Prisma client
+    """Narrow the untyped Prisma client down to the raw-query call this module makes, pinned to the writer."""
+    db: Final = AccessGroupRepository(prisma_client).prisma_client.db  # pyright: ignore[reportAny]  # untyped Prisma client
+    return WriterPinnedClient(db).db  # pyright: ignore[reportAny, reportReturnType]  # untyped Prisma client behind the pin
 
 
 async def _invalidate_access_group_cache(access_group_id: str) -> None:
diff --git a/tests/test_litellm/proxy/management_helpers/test_access_group_key_sync.py b/tests/test_litellm/proxy/management_helpers/test_access_group_key_sync.py
new file mode 100644
index 00000000000..60c36e33e09
--- /dev/null
+++ b/tests/test_litellm/proxy/management_helpers/test_access_group_key_sync.py
@@ -0,0 +1,57 @@
+from types import SimpleNamespace
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+
+from litellm.proxy.db.prisma_client import PrismaWrapper
+from litellm.proxy.db.routing_prisma_wrapper import RoutingPrismaWrapper
+from litellm.proxy.management_helpers.access_group_key_sync import (
+    sync_key_access_group_membership,
+    sync_key_regeneration_access_group_membership,
+)
+
+
+def _routed_prisma_client():
+    writer_inner = MagicMock(name="writer_prisma")
+    reader_inner = MagicMock(name="reader_prisma")
+    writer_inner.query_raw = AsyncMock(return_value=[])
+    reader_inner.query_raw = AsyncMock(return_value=[])
+    writer = PrismaWrapper(original_prisma=writer_inner, iam_token_db_auth=False)
+    reader = PrismaWrapper(original_prisma=reader_inner, iam_token_db_auth=False)
+    routing = RoutingPrismaWrapper(writer=writer, reader=reader)
+    return SimpleNamespace(db=routing), writer_inner, reader_inner
+
+
+@pytest.mark.asyncio
+async def test_regeneration_repoint_update_runs_on_the_writer():
+    prisma_client, writer_inner, reader_inner = _routed_prisma_client()
+
+    await sync_key_regeneration_access_group_membership(
+        prisma_client=prisma_client,
+        previous_key_token="old-token",
+        new_key_token="new-token",
+        data=None,
+        existing_key_row=MagicMock(),
+    )
+
+    writer_inner.query_raw.assert_awaited_once()
+    assert writer_inner.query_raw.await_args.args[0].startswith('UPDATE "LiteLLM_AccessGroupTable"')
+    reader_inner.query_raw.assert_not_awaited()
+
+
+@pytest.mark.asyncio
+async def test_membership_attach_and_detach_updates_run_on_the_writer():
+    prisma_client, writer_inner, reader_inner = _routed_prisma_client()
+
+    await sync_key_access_group_membership(
+        prisma_client=prisma_client,
+        key_token="token",
+        previous_access_group_ids=["ag-old"],
+        updated_access_group_ids=["ag-new"],
+    )
+
+    assert writer_inner.query_raw.await_count == 2
+    assert all(
+        call.args[0].startswith('UPDATE "LiteLLM_AccessGroupTable"') for call in writer_inner.query_raw.await_args_list
+    )
+    reader_inner.query_raw.assert_not_awaited()