[Bug]: Gemini model doesn't support audio parameters despite being a TTS model (gemini-2.5-pro-preview-tts)

[ShivamB25]

What happened?

A bug happened!
When attempting to use the gemini-2.5-pro-preview-tts model for audio generation through LiteLLM, the library throws an UnsupportedParamsError stating that "gemini does not support parameters: ['audio']". This occurs even though the model name suggests it's a text-to-speech model that should support audio output.

The error prevents any audio generation functionality from working with Gemini TTS models, despite the model being specifically designed for text-to-speech tasks.

what i expected:
Expected the gemini-2.5-pro-preview-tts model to successfully process audio generation requests since:

1. The model name explicitly includes "tts" (text-to-speech)
2. It should support audio output parameters for TTS functionality
3. LiteLLM should properly handle audio parameters for TTS-capable models

The completion should have generated audio output without throwing an UnsupportedParamsError.

Relevant log output

python3 litellm_main.py
Sending text input to the model for audio generation...
An error occurred during LiteLLM completion: litellm.UnsupportedParamsError: gemini does not support parameters: ['audio'], for model=gemini-2.5-pro-preview-tts. To drop these, set `litellm.drop_params=True` or for proxy:

`litellm_settings:
 drop_params: true`
. 
 If you want to use these params dynamically send allowed_openai_params=['audio'] in your request.
Traceback (most recent call last):
  File "/Users/shivambansal/Downloads/audio-tests/litellm_main.py", line 17, in main
    response = await completion(
                     ^^^^^^^^^^^
  File "/Users/shivambansal/Downloads/audio-tests/.venv/lib/python3.12/site-packages/litellm/utils.py", line 1282, in wrapper
    raise e
  File "/Users/shivambansal/Downloads/audio-tests/.venv/lib/python3.12/site-packages/litellm/utils.py", line 1160, in wrapper
    result = original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/shivambansal/Downloads/audio-tests/.venv/lib/python3.12/site-packages/litellm/main.py", line 3229, in completion
    raise exception_type(
  File "/Users/shivambansal/Downloads/audio-tests/.venv/lib/python3.12/site-packages/litellm/main.py", line 1143, in completion
    optional_params = get_optional_params(
                      ^^^^^^^^^^^^^^^^^^^^
  File "/Users/shivambansal/Downloads/audio-tests/.venv/lib/python3.12/site-packages/litellm/utils.py", line 2977, in get_optional_params
    _check_valid_arg(
  File "/Users/shivambansal/Downloads/audio-tests/.venv/lib/python3.12/site-packages/litellm/utils.py", line 2960, in _check_valid_arg
    raise UnsupportedParamsError(
litellm.exceptions.UnsupportedParamsError: litellm.UnsupportedParamsError: gemini does not support parameters: ['audio'], for model=gemini-2.5-pro-preview-tts. To drop these, set `litellm.drop_params=True` or for proxy:

`litellm_settings:
 drop_params: true`
. 
 If you want to use these params dynamically send allowed_openai_params=['audio'] in your request.

Additional verification:
python3 -c "import litellm; print('Gemini audio output support:', litellm.supports_audio_output('gemini/gemini-2.5-pro-flash-tts'))"
Gemini audio output support: False

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

main-v1.70.2.dev5

Twitter / LinkedIn details

No response

[taelid]

Also encountering this error in litellm proxy v1.17.1, is there any workaround to get Gemini tts working with litellm?

I'm getting the same error with gemini-2.5-flash-preview-tts.

[ShivamB25]

@krrishdholakia still the same erro. via ai studio

curl -X POST \
  https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-pro-preview-tts:generateContent?key=zxxx \
  -H 'Content-Type: ap****on' \
  -H 'Content-Type: application/json' \
  -d '{
  {

    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "What's 1 + 1?"
          }
        ]
      }
    ],
    "generationConfig": {}
  }
  }'```

```litellm.BadRequestError: VertexAIException BadRequestError - {
  "error": {
    "code": 400,
    "message": "The requested combination of response modalities is not supported by the model. models/gemini-2.5-pro-preview-tts accepts the following combination of response modalities:\n* AUDIO\n",
    "status": "INVALID_ARGUMENT"
  }
}

stack trace: Traceback (most recent call last):
  File "/usr/lib/python3.13/site-packages/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py", line 1494, in async_completion
    response = await client.post(
               ^^^^^^^^^^^^^^^^^^
        api_base, headers=headers, json=cast(dict, request_body)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    )  # type: ignore
    ^
  File "/usr/lib/python3.13/site-packages/litellm/litellm_core_utils/logging_utils.py", line 135, in async_wrapper
    result = await func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/llms/custom_httpx/http_handler.py", line 276, in post
    raise e
  File "/usr/lib/python3.13/site-packages/litellm/llms/custom_httpx/http_handler.py", line 232, in post
    response.raise_for_status()
    ~~~~~~~~~~~~~~~~~~~~~~~~~^^
  File "/usr/lib/python3.13/site-packages/httpx/_models.py", line 761, in raise_for_status
    raise HTTPStatusErro```


```VertexAIException BadRequestError - { "error": { "code": 400, "message": "The requested combination of response modalities is not supported by the model. models/gemini-2.5-pro-preview-tts accepts the following combination of response modalities:\n* AUDIO\n", "status": "INVALID_ARGUMENT" } }

using this: https://github.com/berriai/litellm/pkgs/container/litellm/428062442?tag=litellm_stable_release_branch-latest

[AyrennC]

You need to use ['audio'] modalities, here is a sample curl request

curl -X POST --location "http://localhost:4000/v1/chat/completions" \
    -H "Authorization: Bearer sk-xxxx" \
    -H "Content-Type: application/json" \
    -d '{
          "model": "gemini/gemini-2.5-flash-preview-tts",
          "messages": [
            {
              "role": "user",
              "content": "Hellow, world"
            }
          ],
          "modalities": [
            "audio"
          ],
          "audio": {
            "voice": "Kore",
            "format": "pcm16"
          }
        }'