From b68fcee3caeaa0e5be3891d6539bfeddaf9ab866 Mon Sep 17 00:00:00 2001
From: Mateo Wang <277851410+mateo-berri@users.noreply.github.com>
Date: Thu, 3 Sep 2026 14:00:59 -0700
Subject: [PATCH 01/10] fix: treat gpt-6 names as the gpt-5 request family in
 OpenAI and Azure configs (#39631)

Adapted for stable/1.100.x: import-context conflict only (upstream's neighbouring custom_tools import is not on this line); the added and removed lines are identical to upstream.

(cherry picked from commit 025a3ca42fa81017e961815ef229bd04ff9e903a)
---
 .../llms/azure/chat/gpt_5_transformation.py   | 22 ++--------------
 litellm/llms/azure/chat/gpt_transformation.py |  3 ++-
 .../llms/openai/chat/gpt_5_transformation.py  | 26 ++++++++-----------
 .../llms/openai/responses/transformation.py   |  3 ++-
 .../chat/test_azure_gpt5_transformation.py    | 12 +++++++++
 .../test_openai_responses_transformation.py   |  2 ++
 .../llms/openai/test_gpt5_transformation.py   | 14 ++++++++++
 .../llms/openai/test_is_model_gpt_5_model.py  |  4 +++
 tests/test_litellm/test_main.py               | 13 ++++++++++
 9 files changed, 62 insertions(+), 37 deletions(-)

diff --git a/litellm/llms/azure/chat/gpt_5_transformation.py b/litellm/llms/azure/chat/gpt_5_transformation.py
index 6fdd277a04f..3189f5b57ac 100644
--- a/litellm/llms/azure/chat/gpt_5_transformation.py
+++ b/litellm/llms/azure/chat/gpt_5_transformation.py
@@ -7,6 +7,7 @@ from litellm.exceptions import UnsupportedParamsError
 from litellm.llms.openai.chat.gpt_5_transformation import (
     OpenAIGPT5Config,
     _get_effort_level,
+    is_gpt_reasoning_series_name,
 )
 from litellm.types.llms.openai import AllMessageValues
 
@@ -35,26 +36,7 @@ class AzureOpenAIGPT5Config(AzureOpenAIConfig, OpenAIGPT5Config):
 
     @classmethod
     def is_model_gpt_5_model(cls, model: str) -> bool:
-        """Check if the Azure model string refers to a gpt-5 variant.
-
-        Accepts both explicit gpt-5 model names and the ``gpt5_series/`` prefix
-        used for manual routing.
-        """
-        # The gpt-5-chat* family (gpt-5-chat, gpt-5-chat-latest, gpt-5-chat-2025-08-07,
-        # …) are regular chat models: they support temperature and tool_choice but NOT
-        # reasoning_effort.  They must NOT be routed through the GPT-5 reasoning path.
-        #
-        # Versioned chat models such as gpt-5.3-chat and gpt-5.1-chat ARE reasoning
-        # models and must stay on the GPT-5 path.  The distinguishing feature is that
-        # the gpt-5-chat family has a literal "-chat" immediately after "gpt-5"
-        # (i.e. "gpt-5-chat…"), while versioned chat models interpose a minor version
-        # number (i.e. "gpt-5.<digit>-chat").
-        #
-        # Using a startswith("gpt-5-chat") prefix check on the normalized name (rather
-        # than a substring check) makes this boundary explicit and avoids any ambiguity
-        # if future model names coincidentally contain "gpt-5-chat" as an interior run.
-        _normalized: Final = model.split("/")[-1]  # strip provider prefix, e.g. "azure/"
-        return ("gpt-5" in model and not _normalized.startswith("gpt-5-chat")) or "gpt5_series" in model
+        return is_gpt_reasoning_series_name(model) or "gpt5_series" in model
 
     def get_supported_openai_params(self, model: str) -> list[str]:
         """Get supported parameters for Azure OpenAI GPT-5 models.
diff --git a/litellm/llms/azure/chat/gpt_transformation.py b/litellm/llms/azure/chat/gpt_transformation.py
index 2df4ab731ab..a849ca4e15f 100644
--- a/litellm/llms/azure/chat/gpt_transformation.py
+++ b/litellm/llms/azure/chat/gpt_transformation.py
@@ -11,6 +11,7 @@ from litellm.litellm_core_utils.prompt_templates.factory import (
     convert_to_azure_openai_messages,
 )
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
+from litellm.llms.openai.chat.gpt_5_transformation import GPT_REASONING_SERIES_MARKERS
 from litellm.types.llms.azure import (
     API_VERSION_MONTH_SUPPORTED_RESPONSE_FORMAT,
     API_VERSION_YEAR_SUPPORTED_RESPONSE_FORMAT,
@@ -123,7 +124,7 @@ class AzureOpenAIConfig(BaseConfig):
         name family needs the rename, including the ``gpt-5-chat*`` models that are excluded from
         the reasoning path by https://github.com/BerriAI/litellm/issues/13781.
         """
-        return "gpt-5" in model or "gpt5_series" in model
+        return any(marker in model for marker in GPT_REASONING_SERIES_MARKERS) or "gpt5_series" in model
 
     def _is_response_format_supported_model(self, model: str) -> bool:
         """
diff --git a/litellm/llms/openai/chat/gpt_5_transformation.py b/litellm/llms/openai/chat/gpt_5_transformation.py
index 0223be300b0..b02f953425d 100644
--- a/litellm/llms/openai/chat/gpt_5_transformation.py
+++ b/litellm/llms/openai/chat/gpt_5_transformation.py
@@ -61,6 +61,14 @@ def _get_effort_level(value: str | dict | None) -> str | None:
     return None
 
 
+GPT_REASONING_SERIES_MARKERS: Final = ("gpt-5", "gpt-6")
+
+
+def is_gpt_reasoning_series_name(model: str) -> bool:
+    normalized: Final = model.split("/")[-1]
+    return any(marker in model for marker in GPT_REASONING_SERIES_MARKERS) and not normalized.startswith("gpt-5-chat")
+
+
 class OpenAIGPT5Config(OpenAIGPTConfig):
     """Configuration for gpt-5 models including GPT-5-Codex variants.
 
@@ -73,21 +81,7 @@ class OpenAIGPT5Config(OpenAIGPTConfig):
 
     @classmethod
     def is_model_gpt_5_model(cls, model: str) -> bool:
-        # The gpt-5-chat* family (gpt-5-chat, gpt-5-chat-latest, gpt-5-chat-2025-08-07,
-        # …) are regular chat models: they support temperature and tool_choice but NOT
-        # reasoning_effort.  They must NOT be routed through the GPT-5 reasoning path.
-        #
-        # Versioned chat models such as gpt-5.3-chat and gpt-5.1-chat ARE reasoning
-        # models and must stay on the GPT-5 path.  The distinguishing feature is that
-        # the gpt-5-chat family has a literal "-chat" immediately after "gpt-5"
-        # (i.e. "gpt-5-chat…"), while versioned chat models interpose a minor version
-        # number (i.e. "gpt-5.<digit>-chat").
-        #
-        # Using a startswith("gpt-5-chat") prefix check on the normalized name (rather
-        # than a substring check) makes this boundary explicit and avoids any ambiguity
-        # if future model names coincidentally contain "gpt-5-chat" as an interior run.
-        _normalized: Final = model.split("/")[-1]  # strip provider prefix, e.g. "openai/"
-        return "gpt-5" in model and not _normalized.startswith("gpt-5-chat")
+        return is_gpt_reasoning_series_name(model)
 
     @classmethod
     def is_model_gpt_5_search_model(cls, model: str) -> bool:
@@ -122,6 +116,8 @@ class OpenAIGPT5Config(OpenAIGPTConfig):
     def is_model_gpt_5_4_plus_model(cls, model: str) -> bool:
         """Check if the model is gpt-5.4 or newer (5.4, 5.5, 5.6, etc., including pro)."""
         model_name: Final = model.split("/")[-1]
+        if model_name.startswith("gpt-6"):
+            return True
         if not model_name.startswith("gpt-5."):
             return False
         try:
diff --git a/litellm/llms/openai/responses/transformation.py b/litellm/llms/openai/responses/transformation.py
index eac844a790d..6bc6d42cb66 100644
--- a/litellm/llms/openai/responses/transformation.py
+++ b/litellm/llms/openai/responses/transformation.py
@@ -14,6 +14,7 @@ from litellm.litellm_core_utils.llm_response_utils.convert_dict_to_response impo
 )
 from litellm.litellm_core_utils.url_utils import encode_url_path_segment
 from litellm.llms.base_llm.responses.transformation import BaseResponsesAPIConfig
+from litellm.llms.openai.chat.gpt_5_transformation import is_gpt_reasoning_series_name
 from litellm.secret_managers.main import get_secret_str
 from litellm.types.llms.openai import *
 from litellm.types.responses.main import *
@@ -54,7 +55,7 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
         parts: Final = model.split("/")
         if len(parts) > 1 and parts[0] not in ("openai",):
             return False
-        return "gpt-5" in model and "gpt-5-chat" not in model
+        return is_gpt_reasoning_series_name(model)
 
     @staticmethod
     def _supports_reasoning_effort_none(model: str) -> bool:
diff --git a/tests/test_litellm/llms/azure/chat/test_azure_gpt5_transformation.py b/tests/test_litellm/llms/azure/chat/test_azure_gpt5_transformation.py
index e06cae97283..bd0f16a695b 100644
--- a/tests/test_litellm/llms/azure/chat/test_azure_gpt5_transformation.py
+++ b/tests/test_litellm/llms/azure/chat/test_azure_gpt5_transformation.py
@@ -336,3 +336,15 @@ class TestAzureResolvesTheDeclaredDefaultEffort:
             drop_params=True,
         )
         assert ("temperature" in mapped) is temperature_survives
+
+
+def test_azure_gpt_6_astra_takes_the_reasoning_series_request_shape():
+    params = litellm.get_optional_params(
+        model="gpt-6-astra",
+        custom_llm_provider="azure",
+        max_tokens=100,
+        reasoning_effort="max",
+    )
+    assert params["max_completion_tokens"] == 100
+    assert "max_tokens" not in params
+    assert params["reasoning_effort"] == "max"
diff --git a/tests/test_litellm/llms/openai/responses/test_openai_responses_transformation.py b/tests/test_litellm/llms/openai/responses/test_openai_responses_transformation.py
index 1a90db7c1fe..fafe8e29fe4 100644
--- a/tests/test_litellm/llms/openai/responses/test_openai_responses_transformation.py
+++ b/tests/test_litellm/llms/openai/responses/test_openai_responses_transformation.py
@@ -1613,6 +1613,8 @@ class TestResponsesSurfaceSharesTheEffortRule:
             ("gpt-5.6-sol", None, False),
             ("gpt-5.6-terra", "none", True),
             ("gpt-5.6-terra", "medium", False),
+            ("gpt-6-astra", None, False),
+            ("gpt-6-astra", "low", False),
         ],
     )
     def test_temperature_follows_the_resolved_effort(
diff --git a/tests/test_litellm/llms/openai/test_gpt5_transformation.py b/tests/test_litellm/llms/openai/test_gpt5_transformation.py
index 9c5bd34d59a..c86ce4df2ac 100644
--- a/tests/test_litellm/llms/openai/test_gpt5_transformation.py
+++ b/tests/test_litellm/llms/openai/test_gpt5_transformation.py
@@ -1505,3 +1505,17 @@ class TestACatalogueOlderThanTheCodeDoesNotStripTemperature:
             drop_params=True,
         )
         assert "temperature" not in mapped
+
+
+def test_gpt_6_astra_takes_the_reasoning_series_request_shape():
+    params = litellm.get_optional_params(
+        model="gpt-6-astra",
+        custom_llm_provider="openai",
+        max_tokens=100,
+        reasoning_effort="max",
+        verbosity="low",
+    )
+    assert params["max_completion_tokens"] == 100
+    assert "max_tokens" not in params
+    assert params["reasoning_effort"] == "max"
+    assert params["verbosity"] == "low"
diff --git a/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py b/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py
index 1095819c98c..107a1afb2c6 100644
--- a/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py
+++ b/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py
@@ -41,6 +41,8 @@ from litellm.llms.azure.chat.gpt_5_transformation import AzureOpenAIGPT5Config
 
 # Models that MUST be classified as GPT-5 (routed through GPT-5 reasoning path)
 GPT5_MODELS = [
+    "gpt-6-astra",
+    "openai/gpt-6-astra",
     "gpt-5",
     "gpt-5.1",
     "gpt-5.2",
@@ -120,6 +122,8 @@ class TestOpenAIGPT5ConfigIsModelGpt5Model:
 # /v1/responses bridge (when reasoning_effort is set and tools are passed) on
 # is_model_gpt_5_4_plus_model, so the gpt-5.6 family must land on the True side.
 GPT5_4_PLUS_MODELS = [
+    "gpt-6-astra",
+    "openai/gpt-6-astra",
     "gpt-5.4",
     "gpt-5.5",
     "gpt-5.5-pro",
diff --git a/tests/test_litellm/test_main.py b/tests/test_litellm/test_main.py
index 8cf878d05d9..aeda7f74405 100644
--- a/tests/test_litellm/test_main.py
+++ b/tests/test_litellm/test_main.py
@@ -784,6 +784,19 @@ def test_responses_api_bridge_check_gpt_5_4_tools_plus_reasoning_routes_to_respo
     assert model_info.get("mode") == "responses"
 
 
+def test_responses_api_bridge_check_gpt_6_astra_tools_with_default_reasoning_routes_to_responses():
+    from litellm.main import responses_api_bridge_check
+
+    model_info, model = responses_api_bridge_check(
+        model="gpt-6-astra",
+        custom_llm_provider="openai",
+        tools=[{"type": "function", "function": {"name": "get_capital"}}],
+    )
+
+    assert model == "gpt-6-astra"
+    assert model_info.get("mode") == "responses"
+
+
 def test_responses_api_bridge_check_gpt_5_5_tools_plus_reasoning_routes_to_responses():
     """gpt-5.5+ with both tools and reasoning_effort should route to Responses API."""
     from litellm.main import responses_api_bridge_check

From bc9712f04523be986be589a8c3580f878de95f8e Mon Sep 17 00:00:00 2001
From: amasen02 <amasen02@users.noreply.github.com>
Date: Fri, 4 Sep 2026 15:37:53 +0530
Subject: [PATCH 02/10] fix(proxy): invalidate end-user spend counter and cache
 on budget reset (#39726)

Adapted for stable/1.100.x: reflowed one generator to this line's ruff format (upstream reformatted it in a later style commit).

Signed-off-by: amasen02 <amasen02@users.noreply.github.com>
(cherry picked from commit daced81f20a6b98c01beecf66fa997310d90bfb4)
---
 .../proxy/common_utils/reset_budget_job.py    | 21 ++++++++++++++-
 .../common_utils/test_reset_budget_job.py     | 26 +++++++++++++++++++
 2 files changed, 46 insertions(+), 1 deletion(-)

diff --git a/litellm/proxy/common_utils/reset_budget_job.py b/litellm/proxy/common_utils/reset_budget_job.py
index 1682cf12f4e..3d5c6e4a131 100644
--- a/litellm/proxy/common_utils/reset_budget_job.py
+++ b/litellm/proxy/common_utils/reset_budget_job.py
@@ -36,6 +36,7 @@ from litellm.proxy.common_utils.timezone_utils import (
     get_budget_reset_settings,
 )
 from litellm.proxy.common_utils.user_api_key_cache import (
+    end_user_cache_key,
     model_access_group_cache_key,
     model_access_group_spend_counter_key,
     tag_cache_key,
@@ -175,6 +176,21 @@ def _model_access_group_cache_keys(row: _ModelAccessGroupRow) -> tuple[str, ...]
     return (model_access_group_cache_key(row.access_group_name),)
 
 
+def _enduser_counter_key(row: _EndUserRow) -> str:
+    return f"spend:end_user:{row.user_id}"
+
+
+def _enduser_cache_keys(row: _EndUserRow) -> tuple[str, ...]:
+    return (end_user_cache_key(row.user_id),)
+
+
+def _enduser_carried_spend(row: _EndUserRow, caps: Mapping[str, float]) -> float:
+    if not caps:
+        return 0.0
+    effective_budget_id = row.budget_id or litellm.max_end_user_budget_id
+    return _carried_spend(row.spend, caps.get(effective_budget_id) if effective_budget_id is not None else None)
+
+
 def _budget_link_where(
     budget_ids: Sequence[str],
     extra: Mapping[str, object] = MappingProxyType({}),
@@ -648,6 +664,7 @@ class ResetBudgetJob:
             if _rollover_enabled()
             else {}  # mutable-ok: empty sentinel immediately frozen by MappingProxyType
         )
+        endusers: Final[tuple[_EndUserRow, ...]] = await self._collect_endusers_to_reset(budget_ids)
         return _BudgetCascade(
             budgets=tuple(budgets_to_reset),
             budget_ids=budget_ids,
@@ -659,7 +676,7 @@ class ResetBudgetJob:
                 for b in budgets_to_reset
                 if b.budget_id is not None and b.budget_duration is not None
             ),
-            endusers=await self._collect_endusers_to_reset(budget_ids),
+            endusers=endusers,
             counter_resets=(
                 *(
                     (_team_membership_counter_key(row), _row_carried_spend(row, rollover_caps))
@@ -672,6 +689,7 @@ class ResetBudgetJob:
                     (_model_access_group_counter_key(row), _row_carried_spend(row, rollover_caps))
                     for row in model_access_groups
                 ),
+                *((_enduser_counter_key(row), _enduser_carried_spend(row, rollover_caps)) for row in endusers),
             ),
             rollover_caps=rollover_caps,
             cache_keys=(
@@ -680,6 +698,7 @@ class ResetBudgetJob:
                 *(key for row in orgs for key in _org_cache_keys(row)),
                 *(key for row in tags for key in _tag_cache_keys(row)),
                 *(key for row in model_access_groups for key in _model_access_group_cache_keys(row)),
+                *(key for row in endusers for key in _enduser_cache_keys(row)),
             ),
         )
 
diff --git a/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py b/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py
index 03b05bd9d87..e6bbfd8c7d4 100644
--- a/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py
+++ b/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py
@@ -1495,6 +1495,32 @@ def test_budget_table_reset_invalidates_every_tag_not_just_the_first(reset_budge
     assert deleted == {"tag:tenant-a", "tag:tenant-b", "tag:tenant-c"}
 
 
+def test_budget_table_reset_invalidates_enduser_counter_and_cache(reset_budget_job, mock_prisma_client, monkeypatch):
+    """When an end user's budget resets, its Redis spend counter is zeroed and its management cache is evicted."""
+    counter_cache = _make_counter_invalidation_job(monkeypatch)
+    budget = _budget_row(budget_id="budget-1")
+    mock_prisma_client.data["budget"] = [budget]
+    test_enduser = type(
+        "LiteLLM_EndUserTable",
+        (),
+        {
+            "spend": 20.0,
+            "litellm_budget_table": budget,
+            "budget_id": "budget-1",
+            "user_id": "customer-42",
+        },
+    )
+    mock_prisma_client.data["enduser"] = [test_enduser]
+
+    asyncio.run(reset_budget_job.reset_budget_for_litellm_budget_table())
+
+    counter_cache.in_memory_cache.set_cache.assert_any_call(key="spend:end_user:customer-42", value=0.0, ttl=60)
+    counter_cache.redis_cache.async_set_cache.assert_any_await(key="spend:end_user:customer-42", value=0.0, ttl=60)
+    deleted = {call.kwargs.get("key") for call in counter_cache.user_api_key_cache.async_delete_cache.await_args_list}
+    assert "end_user_id:customer-42" in deleted
+
+
+
 def test_budget_table_reset_commits_even_when_cache_eviction_fails(reset_budget_job, mock_prisma_client, monkeypatch):
     """Eviction runs after the commit, so a broken cache cannot undo the write."""
     counter_cache = _make_counter_invalidation_job(monkeypatch)

From 6b103ed06438958736f2c41bd017025937ae1160 Mon Sep 17 00:00:00 2001
From: ryan-crabbe-berri <ryan@berri.ai>
Date: Thu, 10 Sep 2026 17:31:25 -0700
Subject: [PATCH 03/10] fix(reset_budget_job): reset end users by budget link,
 not by user id (#40639)

Adapted for stable/1.100.x: added Final to the test module's typing import; upstream's test file already imported it.

(cherry picked from commit 8a4fae0e174799ce2d3858a55473be0d6247a2a4)
---
 .../proxy/common_utils/reset_budget_job.py    | 46 ++++++-----
 .../test_proxy_budget_reset.py                |  6 +-
 .../common_utils/test_reset_budget_job.py     | 77 +++++++++++++++----
 3 files changed, 89 insertions(+), 40 deletions(-)

diff --git a/litellm/proxy/common_utils/reset_budget_job.py b/litellm/proxy/common_utils/reset_budget_job.py
index 3d5c6e4a131..603f387be1f 100644
--- a/litellm/proxy/common_utils/reset_budget_job.py
+++ b/litellm/proxy/common_utils/reset_budget_job.py
@@ -222,33 +222,31 @@ def _queue_budget_linked_resets(
 
 
 def _queue_enduser_resets(writes: LinkedSpendResetWrites, cascade: "_BudgetCascade") -> None:
-    """End users are matched by id rather than budget link: rows with no
-    budget_id ride the default budget tier (litellm.max_end_user_budget_id).
-    Zero-before-decrement ordering matters here too (see
-    _queue_budget_linked_resets)."""
-    if not cascade.rollover_caps:
-        if cascade.endusers:
-            writes.queue_spend_zero(
-                where={"user_id": {"in": [row.user_id for row in cascade.endusers]}}
-            )  # mutable-ok: prisma where filter must be a dict
+    """End users reset on the budget link like every other gated table, plus a
+    NULL-budget_id branch: rows created implicitly persist no link and ride the
+    default tier (litellm.max_end_user_budget_id).
+
+    Matching on the link rather than enumerating user ids keeps a statement's
+    bind count proportional to the expiring tiers instead of the customer
+    population, which past ~32,700 dependents exceeds PostgreSQL's per-statement
+    bind ceiling and wedges the cascade permanently (#40564).
+    """
+    _queue_budget_linked_resets(writes, cascade, extra=_SPENT_ROWS_WHERE)
+    default_budget_id: Final = litellm.max_end_user_budget_id
+    if default_budget_id is None or default_budget_id not in cascade.budget_ids:
         return
-    tiered: Final = tuple((row.budget_id or litellm.max_end_user_budget_id, row.user_id) for row in cascade.endusers)
-    for budget_id, cap in cascade.rollover_caps.items():
-        if not (
-            user_ids := [uid for bid, uid in tiered if bid == budget_id]
-        ):  # mutable-ok: prisma "in" filter takes a list
-            continue
+    cap: Final = cascade.rollover_caps.get(default_budget_id)
+    if cap is None:
         writes.queue_spend_zero(
-            where={"user_id": {"in": user_ids}, "spend": {"lte": cap}}
+            where={"budget_id": None, **_SPENT_ROWS_WHERE}
         )  # mutable-ok: prisma where filter must be a dict
-        writes.queue_spend_decrement(
-            where={"user_id": {"in": user_ids}, "spend": {"gt": cap}}, amount=cap
-        )  # mutable-ok: prisma where filter must be a dict
-    plain: Final = [
-        uid for bid, uid in tiered if bid is None or bid not in cascade.rollover_caps
-    ]  # mutable-ok: prisma "in" filter takes a list
-    if plain:
-        writes.queue_spend_zero(where={"user_id": {"in": plain}})  # mutable-ok: prisma where filter must be a dict
+        return
+    writes.queue_spend_zero(
+        where={"budget_id": None, "spend": {"gt": 0, "lte": cap}}
+    )  # mutable-ok: prisma where filter must be a dict
+    writes.queue_spend_decrement(
+        where={"budget_id": None, "spend": {"gt": cap}}, amount=cap
+    )  # mutable-ok: prisma where filter must be a dict
 
 
 @dataclass(frozen=True, slots=True)
diff --git a/tests/litellm_utils_tests/test_proxy_budget_reset.py b/tests/litellm_utils_tests/test_proxy_budget_reset.py
index 9f6e1f4c3f7..4103536950d 100644
--- a/tests/litellm_utils_tests/test_proxy_budget_reset.py
+++ b/tests/litellm_utils_tests/test_proxy_budget_reset.py
@@ -401,7 +401,7 @@ async def test_reset_budget_endusers_are_zeroed_with_the_budget_window_advance()
 
     enduser_writes = [c for c in batch_calls if c["table"] == "enduser"]
     assert len(enduser_writes) == 1
-    assert enduser_writes[0]["where"]["user_id"]["in"] == [f"user{i}" for i in range(1, 7)]
+    assert enduser_writes[0]["where"] == {"budget_id": {"in": ["budget1"]}, "spend": {"gt": 0}}
     assert enduser_writes[0]["data"] == {"spend": 0}
 
     budget_writes = [c for c in batch_calls if c["table"] == "budget"]
@@ -602,7 +602,7 @@ async def test_reset_budget_continues_other_categories_on_failure():
     assert len([c for c in batch_calls if c["table"] == "team_membership"]) == 1
     enduser_writes = [c for c in batch_calls if c["table"] == "enduser"]
     assert len(enduser_writes) == 1
-    assert enduser_writes[0]["where"] == {"user_id": {"in": ["user1"]}}
+    assert enduser_writes[0]["where"] == {"budget_id": {"in": ["budget1"]}, "spend": {"gt": 0}}
     assert enduser_writes[0]["data"] == {"spend": 0}
 
     # Check the new batch write path: 2 keys + 1 user (user1 failed) + 2 teams.
@@ -1031,7 +1031,7 @@ async def test_service_logger_endusers_success():
 
     enduser_writes = [c for c in batch_calls if c["table"] == "enduser"]
     assert len(enduser_writes) == 1
-    assert enduser_writes[0]["where"] == {"user_id": {"in": ["user1", "user2"]}}
+    assert enduser_writes[0]["where"] == {"budget_id": {"in": ["budget1"]}, "spend": {"gt": 0}}
 
     proxy_logging_obj.service_logging_obj.async_service_success_hook.assert_called_once()
     (
diff --git a/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py b/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py
index e6bbfd8c7d4..23cf85d9c3a 100644
--- a/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py
+++ b/tests/test_litellm/proxy/common_utils/test_reset_budget_job.py
@@ -4,7 +4,7 @@ import sys
 import types
 from datetime import datetime, timedelta, timezone
 from datetime import time as dt_time
-from typing import Any, Dict, List
+from typing import Any, Dict, Final, List
 from unittest.mock import AsyncMock, MagicMock
 
 import httpx
@@ -403,7 +403,7 @@ def test_reset_budget_for_enduser(reset_budget_job, mock_prisma_client):
         {
             "table": "enduser",
             "op": "update_many",
-            "where": {"user_id": {"in": ["test-enduser-1"]}},
+            "where": {"budget_id": {"in": ["test-budget-1"]}, "spend": {"gt": 0}},
             "data": {"spend": 0},
         }
     ]
@@ -504,7 +504,7 @@ def test_reset_budget_all(reset_budget_job, mock_prisma_client):
         {
             "table": "enduser",
             "op": "update_many",
-            "where": {"user_id": {"in": ["test-enduser-1"]}},
+            "where": {"budget_id": {"in": ["test-budget-1"]}, "spend": {"gt": 0}},
             "data": {"spend": 0},
         }
     ]
@@ -524,6 +524,7 @@ _LINKED_TABLE_CASES = [
     ("org", {"budget_id": {"in": ["7d-budget-tier"]}, "spend": {"gt": 0}}),
     ("tag", {"budget_id": {"in": ["7d-budget-tier"]}, "spend": {"gt": 0}}),
     ("model_access_group", {"budget_id": {"in": ["7d-budget-tier"]}, "spend": {"gt": 0}}),
+    ("enduser", {"budget_id": {"in": ["7d-budget-tier"]}, "spend": {"gt": 0}}),
 ]
 
 
@@ -553,6 +554,48 @@ def test_budget_table_reset_zeroes_spend_on_every_linked_table(
     assert writes[0]["data"] == {"spend": 0}
 
 
+_POSTGRES_MAX_BIND_VARIABLES: Final = 32767
+
+
+def _bind_count(where: Dict[str, Any]) -> int:
+    """Bind variables one prisma where-clause compiles to: each scalar is one
+    placeholder and an ``in`` list contributes one per element."""
+    return sum(len(value["in"]) if isinstance(value, dict) and "in" in value else 1 for value in where.values())
+
+
+@pytest.mark.parametrize("population", [3, 40_000], ids=["small", "over-pg-bind-ceiling"])
+def test_enduser_reset_bind_count_does_not_scale_with_population(reset_budget_job, mock_prisma_client, population):
+    """Regression for #40564.
+
+    Enumerating every dependent user id put one bind variable per customer into
+    a single prepared statement. Past PostgreSQL's ceiling the statement could
+    not be parsed at all, so the whole atomic cascade rolled back,
+    budget_reset_at never advanced, and the tier stayed due on every later tick
+    forever. Matching on the budget link keeps the statement the same size no
+    matter how many customers share a tier.
+    """
+    budget = _budget_row(budget_id="shared-tier", budget_duration="1d")
+    mock_prisma_client.data["budget"] = [budget]
+    mock_prisma_client.data["enduser"] = [
+        types.SimpleNamespace(
+            spend=1.0,
+            litellm_budget_table=budget,
+            user_id=f"cust-{index:08d}",
+            budget_id="shared-tier",
+        )
+        for index in range(population)
+    ]
+
+    asyncio.run(reset_budget_job.reset_budget_for_litellm_budget_table())
+
+    writes = _batch_writes(mock_prisma_client, "enduser")
+    assert [_bind_count(write["where"]) for write in writes] == [2], (
+        f"the cascade must not enumerate {population} user ids: past "
+        f"{_POSTGRES_MAX_BIND_VARIABLES} binds PostgreSQL refuses the statement, got {writes[:1]}"
+    )
+    assert _batch_writes(mock_prisma_client, "budget")[0]["data"]["budget_reset_at"] is not None
+
+
 def test_budget_table_reset_writes_nothing_when_no_budget_is_due(reset_budget_job, mock_prisma_client):
     """Nothing due means no transaction is opened at all."""
     asyncio.run(reset_budget_job.reset_budget_for_litellm_budget_table())
@@ -720,14 +763,22 @@ def test_reset_budget_resets_endusers_with_null_budget_id(reset_budget_job, mock
 
     asyncio.run(reset_budget_job.reset_budget_for_litellm_budget_table())
 
-    # Both end users are zeroed by the same committed statement.
-    enduser_writes = _batch_writes(mock_prisma_client, "enduser")
-    assert len(enduser_writes) == 1, f"Expected a single enduser write, got {enduser_writes}"
-    assert set(enduser_writes[0]["where"]["user_id"]["in"]) == {
-        "enduser-explicit",
-        "enduser-implicit",
-    }
-    assert enduser_writes[0]["data"] == {"spend": 0}
+    # Both end users are zeroed: the linked rows on the tier's budget_id, the
+    # implicit ones on the NULL branch that stands in for the default tier.
+    assert _batch_writes(mock_prisma_client, "enduser") == [
+        {
+            "table": "enduser",
+            "op": "update_many",
+            "where": {"budget_id": {"in": [default_budget_id]}, "spend": {"gt": 0}},
+            "data": {"spend": 0},
+        },
+        {
+            "table": "enduser",
+            "op": "update_many",
+            "where": {"budget_id": None, "spend": {"gt": 0}},
+            "data": {"spend": 0},
+        },
+    ]
 
     # Verify find_many was called to fetch NULL-budget-id end users
     find_many_calls = mock_prisma_client.db.litellm_endusertable.find_many_calls
@@ -3043,13 +3094,13 @@ def test_budget_cascade_carries_enduser_overage_when_rollover_enabled(
     assert {
         "table": "enduser",
         "op": "update_many",
-        "where": {"user_id": {"in": ["enduser-roll"]}, "spend": {"gt": 10.0}},
+        "where": {"budget_id": "budget-roll", "spend": {"gt": 10.0}},
         "data": {"spend": {"decrement": 10.0}},
     } in enduser_writes
     assert {
         "table": "enduser",
         "op": "update_many",
-        "where": {"user_id": {"in": ["enduser-roll"]}, "spend": {"lte": 10.0}},
+        "where": {"budget_id": "budget-roll", "spend": {"gt": 0, "lte": 10.0}},
         "data": {"spend": 0},
     } in enduser_writes
 

From c8bc09ea511dd7930effa40d7c5986339d47b515 Mon Sep 17 00:00:00 2001
From: Yuneng Jiang <yuneng@berri.ai>
Date: Thu, 24 Sep 2026 20:37:38 -0700
Subject: [PATCH 04/10] chore(deps): bump anyio to 4.14.2

---
 uv.lock | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/uv.lock b/uv.lock
index 84b9d1f6426..c774f19300e 100644
--- a/uv.lock
+++ b/uv.lock
@@ -10,7 +10,7 @@ resolution-markers = [
 ]
 
 [options]
-exclude-newer = "2026-08-26T18:33:25.773031Z"
+exclude-newer = "2026-09-22T03:37:36.984495Z"
 exclude-newer-span = "P3D"
 
 [manifest]
@@ -315,16 +315,16 @@ vertex = [
 
 [[package]]
 name = "anyio"
-version = "4.13.0"
+version = "4.14.2"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "exceptiongroup", marker = "python_full_version < '3.11'" },
     { name = "idna" },
     { name = "typing-extensions", marker = "python_full_version < '3.13'" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/19/14/2c5dd9f512b66549ae92767a9c7b330ae88e1932ca57876909410251fe13/anyio-4.13.0.tar.gz", hash = "sha256:334b70e641fd2221c1505b3890c69882fe4a2df910cba14d97019b90b24439dc", size = 231622, upload-time = "2026-03-24T12:59:09.671Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/61/cc/a381afa6efea9f496eff839d4a6a1aed3bfafc7b3ab4b0d1b243a12573dd/anyio-4.14.2.tar.gz", hash = "sha256:cfa139f3ed1a23ee8f88a145ddb5ac7605b8bbfd8592baacd7ce3d8bb4313c7f", size = 260176, upload-time = "2026-07-12T20:29:07.082Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/da/42/e921fccf5015463e32a3cf6ee7f980a6ed0f395ceeaa45060b61d86486c2/anyio-4.13.0-py3-none-any.whl", hash = "sha256:08b310f9e24a9594186fd75b4f73f4a4152069e3853f1ed8bfbf58369f4ad708", size = 114353, upload-time = "2026-03-24T12:59:08.246Z" },
+    { url = "https://files.pythonhosted.org/packages/da/35/f2287558c17e29fafc8ef3daf819bb9834061cfa43bff8014f7df7f63bdc/anyio-4.14.2-py3-none-any.whl", hash = "sha256:9f505dda5ac9f0c8309b5e8bd445a8c2bf7246f3ce950121e45ea15bc41d1494", size = 125813, upload-time = "2026-07-12T20:29:05.763Z" },
 ]
 
 [[package]]

From b830735f565362c3609ee42a1007bd284b241686 Mon Sep 17 00:00:00 2001
From: Yuneng Jiang <yuneng@berri.ai>
Date: Thu, 24 Sep 2026 20:37:38 -0700
Subject: [PATCH 05/10] chore(deps): bump gitpython to 3.1.60

---
 uv.lock | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/uv.lock b/uv.lock
index c774f19300e..b121a2d2976 100644
--- a/uv.lock
+++ b/uv.lock
@@ -10,7 +10,7 @@ resolution-markers = [
 ]
 
 [options]
-exclude-newer = "2026-09-22T03:37:36.984495Z"
+exclude-newer = "2026-09-22T03:37:38.379785Z"
 exclude-newer-span = "P3D"
 
 [manifest]
@@ -2373,14 +2373,14 @@ wheels = [
 
 [[package]]
 name = "gitpython"
-version = "3.1.58"
+version = "3.1.60"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "gitdb" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/26/d6/5f358ff283325580c2003a6d953aea18cfe10ae87b46f5ebc80fa3a386dc/gitpython-3.1.58.tar.gz", hash = "sha256:621416df10ef3fd0e19fabf9172ddeed0fa704d353d04f194eec56a625a95b22", size = 228498, upload-time = "2026-08-04T15:05:49.47Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/84/14/e6b1a48d831755a53c2029351fcef82e70db4a08f338daefe29d8d0cf31c/gitpython-3.1.60.tar.gz", hash = "sha256:e936431879fa85581b4311fa63492ea52251909e2d655b6529c704c904ddcc24", size = 230793, upload-time = "2026-08-25T18:33:46.102Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/ec/0c/9d8752098bc442f0726e64aa6135940b3a96809915d1aa4206c1bb97881d/gitpython-3.1.58-py3-none-any.whl", hash = "sha256:d331e722577f0fd7fc1f857419b3ecc07af66282b933d2a4d95f84a042fdd50f", size = 220183, upload-time = "2026-08-04T15:05:48.025Z" },
+    { url = "https://files.pythonhosted.org/packages/71/63/ba28697918b7c190af9f3f21940d03e8814e25dd4ddd39d6929f3a553995/gitpython-3.1.60-py3-none-any.whl", hash = "sha256:39548bffb8fa0f3a548133348868bb4838e79d73283052207dc97781a569b6b4", size = 221893, upload-time = "2026-08-25T18:33:44.75Z" },
 ]
 
 [[package]]

From c18d0307da93f99660a599c1c4a9903b6c7c79e8 Mon Sep 17 00:00:00 2001
From: Yuneng Jiang <yuneng@berri.ai>
Date: Thu, 24 Sep 2026 20:37:39 -0700
Subject: [PATCH 06/10] chore(deps): bump tornado to 6.5.8

---
 uv.lock | 24 ++++++++++++------------
 1 file changed, 12 insertions(+), 12 deletions(-)

diff --git a/uv.lock b/uv.lock
index b121a2d2976..d33bec1d06c 100644
--- a/uv.lock
+++ b/uv.lock
@@ -10,7 +10,7 @@ resolution-markers = [
 ]
 
 [options]
-exclude-newer = "2026-09-22T03:37:38.379785Z"
+exclude-newer = "2026-09-22T03:37:38.998393Z"
 exclude-newer-span = "P3D"
 
 [manifest]
@@ -9437,19 +9437,19 @@ wheels = [
 
 [[package]]
 name = "tornado"
-version = "6.5.7"
+version = "6.5.8"
 source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/64/24/95ec527ad67b76d59299e5465b3935d05e4294b7e0290a3924b7487df30b/tornado-6.5.7.tar.gz", hash = "sha256:66c513a76cda70d53907bc27cf1447557699c2e95aa48ba27a442ff61c3ddfc2", size = 519252, upload-time = "2026-06-08T17:34:51.232Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/10/d3/343e5bb989d6515b1646cf3d40135d73f3d5e45339bded401b56cdac24dd/tornado-6.5.8.tar.gz", hash = "sha256:9452e1b208a8bd771e2cb1f2ff564985b9b214bdebbe622793e1799e0a6bd23f", size = 520493, upload-time = "2026-08-07T02:12:42.971Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/02/dc/c7043cab6fed8ae159fc1923ce829ada35c4dbd797d408a43858ffaf9639/tornado-6.5.7-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:148b2eb15c2c765a50796172c1e499649b35f30d2e3c3d3e15913cfa56bfb163", size = 448543, upload-time = "2026-06-08T17:34:38.052Z" },
-    { url = "https://files.pythonhosted.org/packages/92/4f/090b1431e5a43df696feceffc268c5383cc079ecb5f08ce58f917109aafe/tornado-6.5.7-cp39-abi3-macosx_10_9_x86_64.whl", hash = "sha256:9da38de27f1da3b78a966f0dae12b5a1ea9afe72ca805d84ff06508272ddf100", size = 446707, upload-time = "2026-06-08T17:34:39.594Z" },
-    { url = "https://files.pythonhosted.org/packages/37/d8/ef374952fd5da67d4463122c2b8e5a96536ec10b4b339254c6dcde81d01c/tornado-6.5.7-cp39-abi3-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:8d759e71906ee783f8867b93bf26a265743da4c1e2f4a018464c1ba019862972", size = 449774, upload-time = "2026-06-08T17:34:41.204Z" },
-    { url = "https://files.pythonhosted.org/packages/35/37/d434c73f4c6e014b745b9b37085f34f40c022f007efff3d7fe65991899f3/tornado-6.5.7-cp39-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8a46347a18f23fb92b396beebe0fb78f61dda0cc302445202c16203d8a18848b", size = 450745, upload-time = "2026-06-08T17:34:42.531Z" },
-    { url = "https://files.pythonhosted.org/packages/b6/2b/56b9aff361d7f1ab728a805ec7d7ea835f8807afa9f5cc690ea0e630efb9/tornado-6.5.7-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:7778b30bef919231265e91c69963ce0f49a1e9c07ac900bbe75b19ce2575ba92", size = 450578, upload-time = "2026-06-08T17:34:43.787Z" },
-    { url = "https://files.pythonhosted.org/packages/02/30/a7444fb23aa76860a14198fab96ac79f1866b0a6e19e26c4381b0938e50f/tornado-6.5.7-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:e726f0c75da7726eec023aa62751ff8878bd2737e34fbdd33b1ae5897d2200f5", size = 449985, upload-time = "2026-06-08T17:34:45.326Z" },
-    { url = "https://files.pythonhosted.org/packages/5c/42/5f0e56c01e8d9d36f4e23f367b85ae6cae0c1ecddd5e6977d8388ad27488/tornado-6.5.7-cp39-abi3-win32.whl", hash = "sha256:f8de3bf12d3efdd0cbe7c8887868198f8a91415e3f29fcf258d9b8eb7b1d9ae4", size = 451047, upload-time = "2026-06-08T17:34:46.784Z" },
-    { url = "https://files.pythonhosted.org/packages/c9/a4/b393076ffb21b469eec5b328a0534cf03a3b90bfc6b1f09507cdd075d938/tornado-6.5.7-cp39-abi3-win_amd64.whl", hash = "sha256:de942f843533a039ef9fa3d9c88c7cd8a7c94553fb5ad0154270989b3d99a2c4", size = 451485, upload-time = "2026-06-08T17:34:48.248Z" },
-    { url = "https://files.pythonhosted.org/packages/71/2e/7b1c769803121b809112cf9a00681c472eae1d80e32d7ec0e0bd61d0d0e1/tornado-6.5.7-cp39-abi3-win_arm64.whl", hash = "sha256:ff934fce95643af5f11efdae618eaa73d469dc588641e5c8d19295a0c65c4796", size = 450506, upload-time = "2026-06-08T17:34:49.702Z" },
+    { url = "https://files.pythonhosted.org/packages/f2/d5/007086fd8df5489338e204f65adce33fd4f21a4999dbb2b9cff2f897b5f4/tornado-6.5.8-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:cc6aa787d7cfab7c3d35189dc7a56fbd2399a569624c730c6b55b3d6531d0403", size = 449487, upload-time = "2026-08-07T02:12:28.682Z" },
+    { url = "https://files.pythonhosted.org/packages/70/c8/5a24a99495903f594f6a199dd7beead1cbc0a13e2cb9102727bcaaf2a997/tornado-6.5.8-cp39-abi3-macosx_10_9_x86_64.whl", hash = "sha256:9715b5eb79735b2bcd454ce216a9275b7c0470e64ea1bf5742f78b2f72b26eeb", size = 447649, upload-time = "2026-08-07T02:12:30.306Z" },
+    { url = "https://files.pythonhosted.org/packages/6e/de/f2e733f386b85962d1b1dc82cd63d169b5b4580062b35397eac9244a41fe/tornado-6.5.8-cp39-abi3-manylinux1_x86_64.manylinux_2_28_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:547d63f450d570c14fe0e8db2cfb14c9bbd1c2503b4a6612586267955aa47b58", size = 450707, upload-time = "2026-08-07T02:12:31.95Z" },
+    { url = "https://files.pythonhosted.org/packages/0b/94/20efeee9a01c141e9ac47c397f81679dfda24b32768fc4fff24e76d36c2c/tornado-6.5.8-cp39-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:7e2360a0ffbe145eca8af0b19cb7203d79b1a98dd4cccdd6b368f6f49c2e3808", size = 451677, upload-time = "2026-08-07T02:12:33.512Z" },
+    { url = "https://files.pythonhosted.org/packages/42/ec/a96ccb8ccf0de2b7bc2c5fa1608a4803735018242e90c4882365a9fd418f/tornado-6.5.8-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:5d242290bdf7ab3151bc1065fdd75c0dcc21cbc7b49f22a4c56329c2d6566d22", size = 451510, upload-time = "2026-08-07T02:12:35.346Z" },
+    { url = "https://files.pythonhosted.org/packages/29/b5/93185859245ad3f00e62175f29607346788b696369347f0146e0421286bb/tornado-6.5.8-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:7b94ff0e128fe0542f3bd331fb44d06260fc4ac16881545159f34ef08aad4195", size = 450917, upload-time = "2026-08-07T02:12:36.963Z" },
+    { url = "https://files.pythonhosted.org/packages/97/cf/fe33cf062834487d34d1559746a4a12521033c22645b6d74d4bca702e018/tornado-6.5.8-cp39-abi3-win32.whl", hash = "sha256:67832909c4779c64942380cb5f044a5c6163d00831472d80e25e115de9917836", size = 451952, upload-time = "2026-08-07T02:12:38.512Z" },
+    { url = "https://files.pythonhosted.org/packages/cb/e1/468ad54333e92ccb62627e62cb88e5fc14a2171daa67ed47b1b8542d5b86/tornado-6.5.8-cp39-abi3-win_amd64.whl", hash = "sha256:11881db6b7c168494be2c2d12e65931451bdf7ee718535418ae1d8855dd5a0ee", size = 452391, upload-time = "2026-08-07T02:12:39.971Z" },
+    { url = "https://files.pythonhosted.org/packages/ad/3e/cd5e4f06e34cde33b8ef66cf36aa2b5ad46354cc1af7d2136bbe365fee1d/tornado-6.5.8-cp39-abi3-win_arm64.whl", hash = "sha256:68a7468c7e289f8514d7d664101753903217eff1bb6822c6b5994a0b5f5bcb26", size = 451411, upload-time = "2026-08-07T02:12:41.469Z" },
 ]
 
 [[package]]

From 559dce5c83f911dceb3397c09cf88f68c9330411 Mon Sep 17 00:00:00 2001
From: Yuneng Jiang <yuneng@berri.ai>
Date: Thu, 24 Sep 2026 20:37:40 -0700
Subject: [PATCH 07/10] chore(deps): bump pypdf to 6.16.1

---
 uv.lock | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/uv.lock b/uv.lock
index d33bec1d06c..131f4d03c22 100644
--- a/uv.lock
+++ b/uv.lock
@@ -10,7 +10,7 @@ resolution-markers = [
 ]
 
 [options]
-exclude-newer = "2026-09-22T03:37:38.998393Z"
+exclude-newer = "2026-09-22T03:37:39.620125Z"
 exclude-newer-span = "P3D"
 
 [manifest]
@@ -7560,14 +7560,14 @@ wheels = [
 
 [[package]]
 name = "pypdf"
-version = "6.15.0"
+version = "6.16.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "typing-extensions", marker = "python_full_version < '3.11'" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/17/17/ee75a92718ec7212de831e71454d702225aa5e474a805cce169806044453/pypdf-6.15.0.tar.gz", hash = "sha256:d39c4d955a76409284a905e2d65b40076d77ab76129e0faaeeb6612403ecfc79", size = 6993794, upload-time = "2026-08-06T13:06:49.929Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/b6/5a/df92d1c1ef8806ca28f20f978ee059894868d93de797a7e2edebe7fe1a43/pypdf-6.16.1.tar.gz", hash = "sha256:c4d1b43ddae921387321cf63936cd16a7743b91d2da92f165c149a195c972ba9", size = 7003737, upload-time = "2026-08-14T12:24:04.531Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/af/72/ce3067ac31e214a66388159f8462ddb8c13dd00170f24d555a1f1ae8ee91/pypdf-6.15.0-py3-none-any.whl", hash = "sha256:14e001d6504822cb1ca9c7ed9a69bccb320f59b320730f55af804361abe4d5ee", size = 378123, upload-time = "2026-08-06T13:06:47.709Z" },
+    { url = "https://files.pythonhosted.org/packages/33/a1/724b18d6757ab7253a8fecd3a430eb8d980ed26872ba16651e7b5ddfc63f/pypdf-6.16.1-py3-none-any.whl", hash = "sha256:63fec31c4092ae50b6729beedcb469055b60d20c834bde1c402df241f371f644", size = 382924, upload-time = "2026-08-14T12:24:02.854Z" },
 ]
 
 [[package]]

From 78be1b73031e0de15bce26ac35941da9a1b40fd0 Mon Sep 17 00:00:00 2001
From: Yuneng Jiang <yuneng@berri.ai>
Date: Thu, 24 Sep 2026 20:37:40 -0700
Subject: [PATCH 08/10] chore(deps): bump soupsieve to 2.9

---
 uv.lock | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/uv.lock b/uv.lock
index 131f4d03c22..dfad728e891 100644
--- a/uv.lock
+++ b/uv.lock
@@ -10,7 +10,7 @@ resolution-markers = [
 ]
 
 [options]
-exclude-newer = "2026-09-22T03:37:39.620125Z"
+exclude-newer = "2026-09-22T03:37:40.238845Z"
 exclude-newer-span = "P3D"
 
 [manifest]
@@ -8959,11 +8959,11 @@ wheels = [
 
 [[package]]
 name = "soupsieve"
-version = "2.8.4"
+version = "2.9"
 source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/47/2c/0a5f6f8ee0d5589e48c7640213ed5175d52cf540a06725b628cc1a45d6ce/soupsieve-2.8.4.tar.gz", hash = "sha256:e121fd02e975c695e4e9e8774a5ee35d74714b59307868dcc5319ad2d9e3328e", size = 121110, upload-time = "2026-05-24T13:55:57.154Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/80/f1/93422647dd7e461f23d254e6b2bfa687a85b53aeb4903fcdbb74474d4584/soupsieve-2.9.tar.gz", hash = "sha256:acee8417325c5653e1377dc31eccad59eb82cbc65942afe6174c53b3aaad63fc", size = 122122, upload-time = "2026-07-19T01:35:18.425Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/5e/f5/0c41cb68dcae6b7de4fac4188a3a9589e21fb31df21ea3a2e888db95e6c9/soupsieve-2.8.4-py3-none-any.whl", hash = "sha256:e7e6b0769c8f51ed59acab6e994b00621096cfb1c640a7509295987388fbaf65", size = 37304, upload-time = "2026-05-24T13:55:55.406Z" },
+    { url = "https://files.pythonhosted.org/packages/7b/d6/3185ab5ad1280319b31986898f3206dd7227cd75e293d4dba2a5e6bf27a0/soupsieve-2.9-py3-none-any.whl", hash = "sha256:a2b2c76d67df2382d245409fd71e321a571717e58463efa32ace87dcadac2c12", size = 37387, upload-time = "2026-07-19T01:35:17.106Z" },
 ]
 
 [[package]]

From 5651c5a008c0477ac5de2a3c8a937e9d82cdc5e3 Mon Sep 17 00:00:00 2001
From: Yuneng Jiang <yuneng@berri.ai>
Date: Thu, 24 Sep 2026 20:39:38 -0700
Subject: [PATCH 09/10] =?UTF-8?q?bump:=20version=201.100.2=20=E2=86=92=201?=
 =?UTF-8?q?.100.3?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 pyproject.toml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/pyproject.toml b/pyproject.toml
index 51a60f8fa7b..780e8eda5ed 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "litellm"
-version = "1.100.2"
+version = "1.100.3"
 description = "Library to easily interface with LLM API providers"
 readme = "README.md"
 requires-python = ">=3.10, <3.15"
@@ -311,7 +311,7 @@ members = ["enterprise", "litellm-proxy-extras"]
 profile = "black"
 
 [tool.commitizen]
-version = "1.100.2"
+version = "1.100.3"
 version_files = [
     "pyproject.toml:^version",
 ]

From 04fcee8ff122ea985755336cbe9302cf33351ea1 Mon Sep 17 00:00:00 2001
From: Yuneng Jiang <yuneng@berri.ai>
Date: Thu, 24 Sep 2026 20:39:38 -0700
Subject: [PATCH 10/10] chore: refresh uv.lock for 1.100.3

---
 uv.lock | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/uv.lock b/uv.lock
index dfad728e891..54d5934b64d 100644
--- a/uv.lock
+++ b/uv.lock
@@ -10,7 +10,7 @@ resolution-markers = [
 ]
 
 [options]
-exclude-newer = "2026-09-22T03:37:40.238845Z"
+exclude-newer = "2026-09-22T03:39:38.223725Z"
 exclude-newer-span = "P3D"
 
 [manifest]
@@ -4266,7 +4266,7 @@ wheels = [
 
 [[package]]
 name = "litellm"
-version = "1.100.2"
+version = "1.100.3"
 source = { editable = "." }
 dependencies = [
     { name = "aiohttp" },