Merge cbdffd27f9 into 0c98afa780

litellm/integrations/websearch_interception/handler.py

@@ -1794,11 +1794,9 @@ class WebSearchInterceptionLogger(CustomLogger):
             k: v for k, v in kwargs.items() if not k.startswith("_websearch_interception") and k not in internal_params
         }
 
-        full_model_name = model
-        if "custom_llm_provider" in kwargs:
-            custom_llm_provider: Final = kwargs["custom_llm_provider"]
-            if not model.startswith(custom_llm_provider) and "/" not in model:
-                full_model_name = f"{custom_llm_provider}/{model}"
+        from litellm.litellm_core_utils.core_helpers import qualify_provider_stripped_model
+
+        full_model_name: Final = qualify_provider_stripped_model(model, kwargs.get("custom_llm_provider", ""))
 
         verbose_logger.debug(
             "WebSearchInterception: Built chat completion request patch model=%s messages=%d",

litellm/litellm_core_utils/core_helpers.py

@@ -29,6 +29,13 @@ def is_codex_user_agent(user_agent: str) -> bool:
     return bool(_CODEX_CLIENT_PREFIX_RE.match(user_agent))
 
 
+def qualify_provider_stripped_model(model: str, custom_llm_provider: str) -> str:
+    """Put the provider prefix back on a provider-stripped model."""
+    if not custom_llm_provider or model.startswith(f"{custom_llm_provider}/"):
+        return model
+    return f"{custom_llm_provider}/{model}"
+
+
 def safe_divide_seconds(seconds: float, denominator: float, default: float | None = None) -> float | None:
     """
     Safely divide seconds by denominator, handling zero division.

litellm/llms/custom_httpx/llm_http_handler.py

@@ -5599,9 +5599,9 @@ class BaseLLMHTTPHandler:
         if patch.messages is None:
             raise ValueError("Agentic loop plan missing patched messages")
 
-        full_model_name = patch.model or model
-        if "/" not in full_model_name:
-            full_model_name = f"{custom_llm_provider}/{full_model_name}"
+        from litellm.litellm_core_utils.core_helpers import qualify_provider_stripped_model
+
+        full_model_name: Final = qualify_provider_stripped_model(patch.model or model, custom_llm_provider)
 
         optional_params_for_followup: Final = dict(optional_params)
         optional_params_for_followup.update(patch.optional_params)

tests/test_litellm/litellm_core_utils/test_core_helpers.py

@@ -11,6 +11,7 @@ from litellm.litellm_core_utils.core_helpers import (
     get_or_create_metadata_bucket,
     map_finish_reason,
     normalize_drop_params,
+    qualify_provider_stripped_model,
     reconstruct_model_name,
     redact_nested_match_and_regex_keys,
 )
@@ -406,3 +407,36 @@ class TestIsExpectedClientError:
             category=RateLimitErrorCategory.VENDOR_RATE_LIMIT,
         )
         assert is_expected_client_error(vendor_limit) is False
+
+
+class TestQualifyProviderStrippedModel:
+    """#38829: a model whose remainder still holds a slash must keep its provider prefix."""
+
+    @pytest.mark.parametrize(
+        "model,provider,expected",
+        [
+            ("mantle/anthropic.claude-sonnet-5", "bedrock", "bedrock/mantle/anthropic.claude-sonnet-5"),
+            ("invoke/anthropic.claude-v2", "bedrock", "bedrock/invoke/anthropic.claude-v2"),
+            ("openai/gpt-4o", "openrouter", "openrouter/openai/gpt-4o"),
+            ("gpt-4o", "openai", "openai/gpt-4o"),
+            ("claude-sonnet-4-5", "anthropic", "anthropic/claude-sonnet-4-5"),
+        ],
+    )
+    def test_the_provider_prefix_is_restored(self, model, provider, expected):
+        assert qualify_provider_stripped_model(model, provider) == expected
+
+    @pytest.mark.parametrize(
+        "model,provider",
+        [
+            ("bedrock/mantle/anthropic.claude-sonnet-5", "bedrock"),
+            ("openai/gpt-4o", "openai"),
+        ],
+    )
+    def test_an_already_qualified_model_is_left_alone(self, model, provider):
+        assert qualify_provider_stripped_model(model, provider) == model
+
+    def test_a_provider_that_only_shares_a_prefix_is_still_qualified(self):
+        assert qualify_provider_stripped_model("openai_like/foo", "openai") == "openai/openai_like/foo"
+
+    def test_no_provider_leaves_the_model_untouched(self):
+        assert qualify_provider_stripped_model("gpt-4o", "") == "gpt-4o"

tests/test_litellm/llms/custom_httpx/test_llm_http_handler.py

@@ -3713,3 +3713,47 @@ def test_image_edit_handler_keeps_the_sync_transform():
     assert config.transform_calls == ["sync"]
     assert captured["body"] == {"transformed_by": "sync"}
     assert response.data[0].b64_json == "sync"
+
+
+class TestAgenticFollowUpKeepsTheProviderPrefix:
+    """#38829: the follow-up must re-dispatch a model litellm.acompletion can route."""
+
+    @staticmethod
+    def _plan():
+        from litellm.types.integrations.custom_logger import (
+            AgenticLoopPlan,
+            AgenticLoopRequestPatch,
+        )
+
+        return AgenticLoopPlan(
+            run_agentic_loop=True,
+            request_patch=AgenticLoopRequestPatch(messages=[{"role": "user", "content": "hi"}]),
+        )
+
+    async def _run_followup(self, model: str, custom_llm_provider: str):
+        from litellm.llms.custom_httpx.llm_http_handler import BaseLLMHTTPHandler
+
+        return await BaseLLMHTTPHandler()._execute_chat_completion_agentic_plan(
+            plan=self._plan(),
+            model=model,
+            messages=[{"role": "user", "content": "hi"}],
+            optional_params={"mock_response": "ok from followup"},
+            kwargs={},
+            custom_llm_provider=custom_llm_provider,
+            depth=0,
+            max_loops=2,
+            fingerprints=[],
+            fingerprint="fp",
+        )
+
+    @pytest.mark.asyncio
+    async def test_a_sub_path_model_still_resolves_a_provider(self):
+        response = await self._run_followup("mantle/anthropic.claude-sonnet-5", "bedrock")
+
+        assert response.choices[0].message.content == "ok from followup"
+
+    @pytest.mark.asyncio
+    async def test_an_ordinary_model_still_resolves_a_provider(self):
+        response = await self._run_followup("gpt-4o-mini", "openai")
+
+        assert response.choices[0].message.content == "ok from followup"