Add support for together.ai image generation

[tombii]

Title

Together.ai will add id, model and object to the response.

Type

🐛 Bug Fix

Changes

Added optional parameters that together.ai sends with it's response.

{'response_object': {'created': None, 'data': [{'b64_json': '/9j/.../9k=', 'revised_prompt': None, 'url': None, 'timings': {'inference': 0.9873751439154148}, 'index': 0}], 'id': '913e3620e9xxxxxx-LOC', 'model': 'black-forest-labs/FLUX.1-schnell-Free', 'object': 'list', 'hidden_params': {'additional_headers': {}}}, 'model_response_object': ImageResponse(created=1739883745, data=[], usage=Usage(completion_tokens=0, prompt_tokens=0, total_tokens=0, completion_tokens_details=None, prompt_tokens_details=None)), 'response_type': 'image_generation', 'stream': False, 'start_time': None, 'end_time': None, 'hidden_params': None, '_response_headers': None, 'convert_tool_call_to_json_mode': None }

[vercel[bot]]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
litellm	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Apr 14, 2025 8:58am

[krrishdholakia]

Hi @tombii is there a reason the existing flow does not work?

[tombii]

@krrishdholakia As you can see below, the response_object contains id, model and object which are not in the init and therefore throws an error. Maybe there are better ways to do this but the provided patch fixes the issue for me.

Here's the error message:

received_args={'response_object': {'created': None, 'data': [{'b64_json': '/9j/.../2Q==', 'revised_prompt': None, 'url': None, 'timings': {'inference': 0.9612685777246952}, 'index': 0}], 'id': '91559891cxxx-PDX', 'model': 'black-forest-labs/FLUX.1-schnell-Free', 'object': 'list', 'hidden_params': {'additional_headers': {}}}, 'model_response_object': ImageResponse(created=1740128949, data=[], usage=Usage(completion_tokens=0, prompt_tokens=0, total_tokens=0, completion_tokens_details=None, prompt_tokens_details=None)), 'response_type': 'image_generation', 'stream': False, 'start_time': None, 'end_time': None, 'hidden_params': None, '_response_headers': None, 'convert_tool_call_to_json_mode': None}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/lib/python3.13/site-packages/litellm/router.py", line 3155, in async_function_with_retries
    response = await self.make_call(original_function, *args, **kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/router.py", line 3271, in make_call
    response = await response
               ^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/router.py", line 1763, in _aimage_generation
    raise e
  File "/usr/lib/python3.13/site-packages/litellm/router.py", line 1750, in _aimage_generation
    response = await response
               ^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 1394, in wrapper_async
    raise e
  File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 1253, in wrapper_async
    result = await original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/main.py", line 4443, in aimage_generation
    raise exception_type(
          ~~~~~~~~~~~~~~^
        model=model,
        ^^^^^^^^^^^^
    ...<3 lines>...
        extra_kwargs=kwargs,
        ^^^^^^^^^^^^^^^^^^^^
    )
    ^
  File "/usr/lib/python3.13/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2207, in exception_type
    raise e  # it's already mapped
    ^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 463, in exception_type
    raise APIConnectionError(
    ...<7 lines>...
    )
litellm.exceptions.APIConnectionError: litellm.APIConnectionError: APIConnectionError: OpenAIException - Invalid response object Traceback (most recent call last):
  File "/usr/lib/python3.13/site-packages/litellm/litellm_core_utils/llm_response_utils/convert_dict_to_response.py", line 586, in convert_to_model_response_object
    return LiteLLMResponseObjectHandler.convert_to_image_response(
           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
        response_object=response_object,
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        model_response_object=model_response_object,
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        hidden_params=hidden_params,
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    )
    ^
  File "/usr/lib/python3.13/site-packages/litellm/litellm_core_utils/llm_response_utils/convert_dict_to_response.py", line 264, in convert_to_image_response
    model_response_object = ImageResponse(**model_response_dict)
TypeError: ImageResponse.__init__() got an unexpected keyword argument 'id'

[CLAassistant]

All committers have signed the CLA.

[Setom29]

I am encountering the same issue with the model black-forest-labs/FLUX.1-dev (“TypeError: ImageResponse.init() got an unexpected keyword argument 'id'”). Could you please advise when these changes might be available in production?

[krrishdholakia]

This is already resolved with the new ImageResponse object - https://github.com/BerriAI/litellm/blob/197e7efa8f097bb935cf86dc4100422487a40955/litellm/types/utils.py#L1606

Tested the input you shared, and i can see it working

[krrishdholakia]

@Setom29 what version are you on? please try v1.74.3 and let me know if that solves it.

[Setom29]

@krrishdholakia Thank you for the suggestion. I was using version 1.69.3, which I had installed not so long ago and expected to be up to date. After updating to v1.74.3, everything is working as expected.