From 46a8025dd69c527d6a2a33bc84cb6484ce1a8ca0 Mon Sep 17 00:00:00 2001
From: Mateo Wang <277851410+mateo-berri@users.noreply.github.com>
Date: Mon, 6 Jul 2026 10:41:49 -0700
Subject: [PATCH] fix(main): forward verbosity param to chat completion
 providers (#32254)

---
 litellm/main.py                 |  2 +
 tests/test_litellm/test_main.py | 87 +++++++++++++++++++++++++++++++++
 2 files changed, 89 insertions(+)

diff --git a/litellm/main.py b/litellm/main.py
index b3c176a1347..2ace46a16fb 100644
--- a/litellm/main.py
+++ b/litellm/main.py
@@ -582,6 +582,7 @@ async def acompletion(
         "api_key": api_key,
         "model_list": model_list,
         "reasoning_effort": reasoning_effort,
+        "verbosity": verbosity,
         "safety_identifier": safety_identifier,
         "service_tier": service_tier,
         "extra_headers": extra_headers,
@@ -5193,6 +5194,7 @@ def completion(  # type: ignore
             "parallel_tool_calls": parallel_tool_calls,
             "messages": messages,
             "reasoning_effort": reasoning_effort,
+            "verbosity": verbosity,
             "thinking": thinking,
             "web_search_options": web_search_options,
             "include_server_side_tool_invocations": (
diff --git a/tests/test_litellm/test_main.py b/tests/test_litellm/test_main.py
index f24df599505..28cf4fa0744 100644
--- a/tests/test_litellm/test_main.py
+++ b/tests/test_litellm/test_main.py
@@ -638,6 +638,93 @@ def test_bedrock_llama():
     )
 
 
+def _mocked_openai_chat_response(model: str) -> httpx.Response:
+    return httpx.Response(
+        status_code=200,
+        json={
+            "id": "chatcmpl-123",
+            "object": "chat.completion",
+            "created": 1677652288,
+            "model": model,
+            "choices": [
+                {
+                    "index": 0,
+                    "message": {
+                        "role": "assistant",
+                        "content": "Hello from mocked response!",
+                    },
+                    "finish_reason": "stop",
+                }
+            ],
+            "usage": {
+                "prompt_tokens": 9,
+                "completion_tokens": 12,
+                "total_tokens": 21,
+            },
+        },
+    )
+
+
+def test_completion_forwards_verbosity_in_raw_request(respx_mock: respx.MockRouter):
+    """Regression test: completion() must forward the verbosity param to the provider request body."""
+    from litellm.types.utils import CallTypes
+    from litellm.utils import return_raw_request
+
+    model = "gpt-5.2"
+    messages = [{"role": "user", "content": "hi"}]
+    respx_mock.post("https://api.openai.com/v1/chat/completions").mock(
+        return_value=_mocked_openai_chat_response(model)
+    )
+
+    request = return_raw_request(
+        endpoint=CallTypes.completion,
+        kwargs={
+            "model": model,
+            "messages": messages,
+            "verbosity": "high",
+        },
+    )
+
+    assert request["raw_request_body"]["verbosity"] == "high"
+    assert request["raw_request_body"]["model"] == model
+    assert request["raw_request_body"]["messages"] == messages
+
+
+@pytest.mark.asyncio
+async def test_acompletion_forwards_verbosity_to_provider_request(
+    respx_mock: respx.MockRouter, monkeypatch
+):
+    """Regression test: acompletion() must forward the verbosity param to the provider request body."""
+    original_disable_aiohttp = litellm.disable_aiohttp_transport
+    try:
+        litellm.disable_aiohttp_transport = True
+        monkeypatch.setenv("DISABLE_AIOHTTP_TRANSPORT", "True")
+        litellm.in_memory_llm_clients_cache.flush_cache()
+
+        model = "gpt-5.2"
+        messages = [{"role": "user", "content": "hi"}]
+        mock_route = respx_mock.post("https://api.openai.com/v1/chat/completions").mock(
+            return_value=_mocked_openai_chat_response(model)
+        )
+
+        response = await litellm.acompletion(
+            model=model,
+            messages=messages,
+            verbosity="low",
+            api_key="fake-openai-api-key",
+        )
+
+        assert response.choices[0].message.content == "Hello from mocked response!"
+        assert mock_route.called
+        request_body = json.loads(respx_mock.calls[0].request.read())
+        assert request_body["verbosity"] == "low"
+        assert request_body["model"] == model
+        assert request_body["messages"] == messages
+    finally:
+        litellm.disable_aiohttp_transport = original_disable_aiohttp
+        litellm.in_memory_llm_clients_cache.flush_cache()
+
+
 def test_responses_api_bridge_check_strips_responses_prefix():
     """Test that responses_api_bridge_check strips 'responses/' prefix and sets mode."""
     from litellm.main import responses_api_bridge_check