fix(anthropic_messages): gate sampling params on /v1/messages like /chat/completions (#37868)

* fix(anthropic_messages): gate sampling params on /v1/messages like /chat/completions

/v1/messages forwarded temperature/top_p/top_k raw to models that removed
sampling params (supports_sampling_params: false — Claude 4.7+/Fable 5),
producing provider 400s that router fallbacks mask as silent model
downgrades. The chat path already gates these via
AnthropicModelInfo._apply_sampling_param; reuse it in
get_requested_anthropic_messages_optional_param so both endpoints agree:
drop under drop_params, else raise the clean client-side 400.

Fixes #35053

* test(anthropic): drive new sampling-param tests off the kwarg, not the global

The five tests added here set `litellm.drop_params = True` under a manual
try/finally. That trips TQ005 (module-global mutation, 10 new violations
over the ceiling) and it leaks process-wide if the finally is ever
skipped, which is what the save/restore conftest exists to paper over.

`get_requested_anthropic_messages_optional_param` already takes
`drop_params` as a kwarg, and that is the path /v1/messages actually
uses, so pass it directly. `monkeypatch.setattr` pins the global to
False so each test proves the per-request flag alone is sufficient and
cannot pass on a leaked global.

Verified: TQ gate clean, all 10 tests pass, and the 3 that assert the
new gating still fail with the fix in utils.py reverted.

---------

Co-authored-by: Mihidum Hettiyahandi <55163074+mihidumh@users.noreply.github.com>

litellm/llms/anthropic/experimental_pass_through/messages/utils.py

@@ -44,6 +44,7 @@ class AnthropicMessagesRequestUtils:
         filtered_params: Final = {k: v for k, v in params.items() if k in valid_keys and v is not None}
         if model is not None:
             from litellm.llms.anthropic.chat.transformation import AnthropicConfig
+            from litellm.llms.anthropic.common_utils import AnthropicModelInfo
 
             AnthropicConfig._maybe_drop_speed_param(
                 model=model,
@@ -51,6 +52,16 @@ class AnthropicMessagesRequestUtils:
                 drop_params=drop_params,
                 custom_llm_provider=custom_llm_provider,
             )
+            for param in ("temperature", "top_p", "top_k"):
+                if param in filtered_params:
+                    AnthropicModelInfo._apply_sampling_param(  # pyright: ignore[reportPrivateUsage]  # same gating the /chat/completions path applies; forking it would drift
+                        optional_params=filtered_params,
+                        model=model,
+                        param=param,
+                        value=filtered_params.pop(param),
+                        drop_params=drop_params,
+                        output_key=param,
+                    )
         return cast(AnthropicMessagesRequestOptionalParams, filtered_params)
 
 

tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_request_optional_param_utils.py

@@ -6,6 +6,8 @@ Regression tests for the /v1/messages request-parse fast paths:
   while resolving the (static) type hints only once per process.
 """
 
+import pytest
+
 import litellm
 from litellm.llms.anthropic.experimental_pass_through.messages.utils import (
     AnthropicMessagesRequestUtils,
@@ -88,3 +90,61 @@ def test_drop_params_keeps_speed_for_supporting_model():
         litellm.drop_params = original
 
     assert result == {"speed": "fast"}
+
+
+def test_drop_params_strips_sampling_params_for_unsupported_model(monkeypatch):
+    # claude-opus-4-7 has supports_sampling_params: false in the model map; the
+    # API 400s on these rather than ignoring them.
+    monkeypatch.setattr(litellm, "drop_params", False)
+    result = AnthropicMessagesRequestUtils.get_requested_anthropic_messages_optional_param(
+        params={"temperature": 0.3, "top_p": 0.9, "top_k": 40, "stream": True},
+        model="claude-opus-4-7",
+        drop_params=True,
+    )
+
+    assert result == {"stream": True}
+
+
+def test_drop_params_strips_sampling_params_for_provider_prefixed_model(monkeypatch):
+    # Vertex-routed ids must resolve the same capability flag.
+    monkeypatch.setattr(litellm, "drop_params", False)
+    result = AnthropicMessagesRequestUtils.get_requested_anthropic_messages_optional_param(
+        params={"temperature": 0.3, "top_p": 0.9, "top_k": 40},
+        model="vertex_ai/claude-opus-4-7",
+        drop_params=True,
+    )
+
+    assert result == {}
+
+
+def test_sampling_params_kept_for_supporting_model(monkeypatch):
+    monkeypatch.setattr(litellm, "drop_params", False)
+    result = AnthropicMessagesRequestUtils.get_requested_anthropic_messages_optional_param(
+        params={"temperature": 0.3, "top_p": 0.9, "top_k": 40},
+        model="claude-sonnet-4-6",
+        drop_params=True,
+    )
+
+    assert result == {"temperature": 0.3, "top_p": 0.9, "top_k": 40}
+
+
+def test_temperature_1_kept_for_unsupported_model(monkeypatch):
+    # temperature=1 is the one value these models still accept.
+    monkeypatch.setattr(litellm, "drop_params", False)
+    result = AnthropicMessagesRequestUtils.get_requested_anthropic_messages_optional_param(
+        params={"temperature": 1},
+        model="claude-opus-4-7",
+        drop_params=True,
+    )
+
+    assert result == {"temperature": 1}
+
+
+def test_sampling_param_raises_clean_400_without_drop_params(monkeypatch):
+    monkeypatch.setattr(litellm, "drop_params", False)
+    with pytest.raises(litellm.utils.UnsupportedParamsError, match="does not support temperature"):
+        AnthropicMessagesRequestUtils.get_requested_anthropic_messages_optional_param(
+            params={"temperature": 0.3},
+            model="claude-opus-4-7",
+            drop_params=False,
+        )