LiteLLM / OpenAI Codex Discussion and Support

[ishaan-jaff]

Starting this thread for support questions using OpenAI codex with LiteLLM

Tutorial using Codex with LiteLLM: https://docs.litellm.ai/docs/tutorials/openai_codex

Steps to Use

1. Install OpenAI Codex CLI:

npm i -g @openai/codex

2. Configure model routing in litellm_config.yaml:

model_list:
 - model_name: o3-mini
   litellm_params:
     model: openai/o3-mini
     api_key: os.environ/OPENAI_API_KEY
 - model_name: gemini-2.0-flash
   litellm_params:
     model: gemini/gemini-2.0-flash
     api_key: os.environ/GEMINI_API_KEY

3. Start LiteLLM Proxy:

litellm --config /path/to/config.yaml

4. Point Codex to LiteLLM Proxy (via Env Variables)

We point codex to litellm as a base url and set the api key to the litellm virtual key

export OPENAI_BASE_URL=http://0.0.0.0:4000
export OPENAI_API_KEY="sk-1234"

5. Run Codex with your preferred model:

codex --model gemini-2.0-flash --full-auto

[nivzx]

Thanks @ishaan-jaff for starting this thread.

Here are the steps I followed (after installing codex with npm) to set up Codex with LiteLLM on Docker(Podman, in my case).

1. Fetched LiteLLM Image

 podman pull ghcr.io/berriai/litellm:main-latest

2. Created a LiteLLM config file

(I defined this in ~/.litellm/config.yaml)

model_list:
  - model_name: deepseek-reasoner
    litellm_params:
      model: deepseek/deepseek-reasoner
      api_key: os.environ/DEEPSEEK_API_KEY
      api_base: os.environ/DEEPSEEK_BASE_URL
  - model_name: deepseek-chat
    litellm_params:
      model: deepseek/deepseek-chat
      api_key: os.environ/DEEPSEEK_API_KEY
      api_base: os.environ/DEEPSEEK_BASE_URL

general_settings:
  master_key: sk-1234

3. Assigned API keys in environment variables in .zshrc

export DEEPSEEK_API_KEY=***
export DEEPSEEK_BASE_URSL=https://api.deepseek.com

4. Point Codex to LiteLLM Proxy (via Env Variables)

export OPENAI_BASE_URL=http://0.0.0.0:4000
export OPENAI_API_KEY="sk-1234"

5. Run LiteLLM with Podman

podman run \
    -v ~/.litellm/config.yaml:/app/config.yaml \
    -p 4000:4000 \
    ghcr.io/berriai/litellm:main-latest \
    --config /app/config.yaml --detailed_debug

6. Started codex with DeepSeek model

codex -m deepseek-chat

It loads the available models properly when I send /models in codex, but when I send a message I get this error in codex

⚠️  OpenAI rejected the request. Error details: Status: 400, Code: 400, Type: unknown, Message: 400 litellm.BadRequestError: Responses API not available for custom_llm_provider=deepseek, model: deepseek-chat. Please verify
    your settings and try again.

And there's this log in LiteLLM proxy debug

14:37:00 - LiteLLM Router:INFO: router.py:2484 - aresponses(model=deepseek-chat) Exception litellm.BadRequestError: Responses API not available for custom_llm_provider=deepseek, model: deepseek-chat
14:37:00 - LiteLLM Proxy:ERROR: common_request_processing.py:332 - litellm.proxy.proxy_server._handle_llm_api_exception(): Exception occured - litellm.BadRequestError: Responses API not available for custom_llm_provider=deepseek, model: deepseek-chat
Traceback (most recent call last):
  File "/usr/lib/python3.13/site-packages/litellm/proxy/response_api_endpoints/endpoints.py", line 56, in responses_api
    return await processor.base_process_llm_request(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    ...<16 lines>...
    )
    ^
  File "/usr/lib/python3.13/site-packages/litellm/proxy/common_request_processing.py", line 244, in base_process_llm_request
    responses = await llm_responses
                ^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/router.py", line 3125, in async_wrapper
    return await self._ageneric_api_call_with_fallbacks(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    ...<2 lines>...
    )
    ^
  File "/usr/lib/python3.13/site-packages/litellm/router.py", line 2489, in _ageneric_api_call_with_fallbacks
    raise e
  File "/usr/lib/python3.13/site-packages/litellm/router.py", line 2476, in _ageneric_api_call_with_fallbacks
    response = await response  # type: ignore
               ^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 1452, in wrapper_async
    raise e
  File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 1313, in wrapper_async
    result = await original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/site-packages/litellm/responses/main.py", line 117, in aresponses
    raise litellm.exception_type(
    ...<5 lines>...
    )
  File "/usr/lib/python3.13/site-packages/litellm/responses/main.py", line 109, in aresponses
    init_response = await loop.run_in_executor(None, func_with_context)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.13/concurrent/futures/thread.py", line 59, in run
    result = self.fn(*self.args, **self.kwargs)
  File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 1247, in wrapper
    raise e
  File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 1125, in wrapper
    result = original_function(*args, **kwargs)
  File "/usr/lib/python3.13/site-packages/litellm/responses/main.py", line 245, in responses
    raise litellm.exception_type(
    ...<5 lines>...
    )
  File "/usr/lib/python3.13/site-packages/litellm/responses/main.py", line 189, in responses
    raise litellm.BadRequestError(
    ...<3 lines>...
    )
litellm.exceptions.BadRequestError: litellm.BadRequestError: Responses API not available for custom_llm_provider=deepseek, model: deepseek-chat
14:37:00 - LiteLLM Proxy:DEBUG: common_request_processing.py:341 - An error occurred: litellm.BadRequestError: Responses API not available for custom_llm_provider=deepseek, model: deepseek-chat None

 Debug this by setting `--debug`, e.g. `litellm --model gpt-3.5-turbo --debug`
INFO:     10.88.0.3:44206 - "POST /responses HTTP/1.1" 400 Bad Request

[ishaan-jaff]

ahh I see the issue, please use the dev release I put out @nivzx ghcr.io/berriai/litellm:main-v1.66.3.dev5

[nivzx]

Oh okay, I'll try that release. Thanks!

[sevospl]

ahh I see the issue, please use the dev release I put out @nivzx ghcr.io/berriai/litellm:main-v1.66.3.dev5

it worked for me, however when i ask codex to explain any codebase to me, i get a 400 invalid argument error:

litellm.BadRequestError: VertexAIException BadRequestError: Please ensure that function call turn comes immediately after a user turn or after a function response turn.

[nivzx]

it worked for me, however when i ask codex to explain any codebase to me, i get a 400 invalid argument error:

litellm.BadRequestError: VertexAIException BadRequestError: Please ensure that function call turn comes immediately after a user turn or after a function response turn.

A similar issue for me. It worked for one message that doesn't require referring the codebase, but then from the next message onwards it throws the following error on that request, and all the requests that follows (event the ones that doesn't require referring the codebase).

litellm.exceptions.BadRequestError: litellm.BadRequestError: DeepseekException - Failed to deserialize the JSON body into the target type: messages[2]: invalid type: sequence, expected a string at line 1 column 5032
16:42:08 - LiteLLM Proxy:DEBUG: common_request_processing.py:341 - An error occurred: litellm.BadRequestError: DeepseekException - Failed to deserialize the JSON body into the target type: messages[2]: invalid type: sequence, expected a string at line 1 column 5032
Model: deepseek-chat
API Base: `https://api.deepseek.com/beta`
Messages: `[{'role': 'system', 'content': 'You are operating as and within the Codex CLI, a terminal-based agen`

 Debug this by setting `--debug`, e.g. `litellm --model gpt-3.5-turbo --debug`
INFO:     10.88.0.8:53910 - "POST /responses HTTP/1.1" 400 Bad Request

But looking at the error message, for me it looks like an issue with parsing the message with the system prompt in codex.

Messages: `[{'role': 'system', 'content': 'You are operating as and within the Codex CLI, a terminal-based agen`