Originally created by @PeterDaveHello on GitHub (Jun 27, 2025).
Check Existing Issues
I have searched the existing issues and discussions.
Problem Description
Some newer or specialized OpenAI models are exclusively available via the /v1/responses endpoint. Attempts to use these models with the existing endpoints (such as /v1/chat/completions) result in the following error:
{"error":{"message":"This model is only supported in v1/responses and not in v1/chat/completions.","type":"invalid_request_error","param":"model","code":null}}
Currently, open-webui only supports the /v1/completions and /v1/chat/completions endpoints of the OpenAI API. Without support for /v1/responses, users cannot utilize the full range of OpenAI's latest models and features.
Desired Solution you'd like
Please add support for the /v1/responses API endpoint to ensure compatibility with all current and future OpenAI models. This enhancement will prevent the above error and allow users to leverage the complete capabilities of the OpenAI platform.
Originally created by @PeterDaveHello on GitHub (Jun 27, 2025).
### Check Existing Issues
- [x] I have searched the existing issues and discussions.
### Problem Description
Some newer or specialized OpenAI models are exclusively available via the /v1/responses endpoint. Attempts to use these models with the existing endpoints (such as /v1/chat/completions) result in the following error:
```json
{
"error": {
"message": "This model is only supported in v1/responses and not in v1/chat/completions.",
"type": "invalid_request_error",
"param": "model",
"code": null
}
}
```
Currently, open-webui only supports the /v1/completions and /v1/chat/completions endpoints of the OpenAI API. Without support for /v1/responses, users cannot utilize the full range of OpenAI's latest models and features.
### Desired Solution you'd like
Please add support for the /v1/responses API endpoint to ensure compatibility with all current and future OpenAI models. This enhancement will prevent the above error and allow users to leverage the complete capabilities of the OpenAI platform.
### Alternatives Considered
_No response_
### Additional Context
- [OpenAI API Reference: Responses](https://platform.openai.com/docs/api-reference/responses)
- [Introducing the Responses API – OpenAI Community](https://community.openai.com/t/introducing-the-responses-api/1140929)
- [Related Error Example – OpenAI Community](https://community.openai.com/t/responses-api-endpoint-reference-documentation-errors-and-issues/1140994)
I have been working on a manifold that brings support for the responses API (as a stop gap until if/when it's supported natively). You can test and monitor progress here. Supports visible reasoning summaries, optimized tool calling, web search, and much more. Working on adding deep research as well.
@jrkropp commented on GitHub (Jun 27, 2025):
I have been working on a manifold that brings support for the responses API (as a stop gap until if/when it's supported natively). You can test and monitor progress [here](https://github.com/jrkropp/open-webui-developer-toolkit/tree/alpha-preview/functions/pipes/openai_responses_manifold). Supports visible reasoning summaries, optimized tool calling, web search, and much more. Working on adding deep research as well.
I have been running into the same issue with the deep-research models. @tjbck tbh: I did not find any discussion about this topic. Maybe I searched for the wrong terms? 404: This model is only supported in v1/responses and not in v1/chat/completions is the error message that the UI shows, no discussion search results for it.
@perelin commented on GitHub (Jul 22, 2025):
I have been running into the same issue with the deep-research models. @tjbck tbh: I did not find any discussion about this topic. Maybe I searched for the wrong terms? `404: This model is only supported in v1/responses and not in v1/chat/completions` is the error message that the UI shows, no discussion search results for it.
I have been running into the same issue with the deep-research models. @tjbck tbh: I did not find any discussion about this topic. Maybe I searched for the wrong terms? 404: This model is only supported in v1/responses and not in v1/chat/completions is the error message that the UI shows, no discussion search results for it.
There's one discussion started around March suggesting adding support of responses API.
@Oatu4396 commented on GitHub (Jul 23, 2025):
> I have been running into the same issue with the deep-research models. [@tjbck](https://github.com/tjbck) tbh: I did not find any discussion about this topic. Maybe I searched for the wrong terms? `404: This model is only supported in v1/responses and not in v1/chat/completions` is the error message that the UI shows, no discussion search results for it.
There's one discussion started around March suggesting adding support of responses API.
https://github.com/open-webui/open-webui/discussions/11874
Just wanted to show my support for this too, I think it's pretty essential. @jrkropp good work on what you've done to support this so far, ive used it and works perfect. Thank you.
@fuzedxb1 commented on GitHub (Aug 9, 2025):
Just wanted to show my support for this too, I think it's pretty essential. @jrkropp good work on what you've done to support this so far, ive used it and works perfect. Thank you.
@wizard-leoliao commented on GitHub (Oct 20, 2025):
Is this issue still unresolved? Almost all OpenAI API models except GPT‑5 are not working properly.
@wizard-leoliao commented on GitHub (Oct 20, 2025):
Is this issue still unresolved? Almost all OpenAI API models except GPT‑5 are not working properly.
Responses API is not supported and probably will not be, because it is not an open standard like the OpenAI completions API.
It's like a proprietary endpoint like Anthropic or Gemini API.
Using one of the many pipes, pipelines or other implementations out there is recommended - alternatively, the use of a middleware like LiteLLM should also do the trick
@Classic298 commented on GitHub (Oct 20, 2025):
Responses API is not supported and probably will not be, because it is not an open standard like the OpenAI completions API.
It's like a proprietary endpoint like Anthropic or Gemini API.
Using one of the many pipes, pipelines or other implementations out there is recommended - alternatively, the use of a middleware like LiteLLM should also do the trick
Responses API is not supported and probably will not be, because it is not an open standard like the OpenAI completions API.
It's like a proprietary endpoint like Anthropic or Gemini API.
Using one of the many pipes, pipelines or other implementations out there is recommended - alternatively, the use of a middleware like LiteLLM should also do the trick
in your opinion, what are the chances that v1/responses is ever available
@verynewtocode commented on GitHub (Nov 5, 2025):
> Responses API is not supported and probably will not be, because it is not an open standard like the OpenAI completions API.
>
> It's like a proprietary endpoint like Anthropic or Gemini API.
>
> Using one of the many pipes, pipelines or other implementations out there is recommended - alternatively, the use of a middleware like LiteLLM should also do the trick
in your opinion, what are the chances that v1/responses is ever available
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @PeterDaveHello on GitHub (Jun 27, 2025).
Check Existing Issues
Problem Description
Some newer or specialized OpenAI models are exclusively available via the /v1/responses endpoint. Attempts to use these models with the existing endpoints (such as /v1/chat/completions) result in the following error:
Currently, open-webui only supports the /v1/completions and /v1/chat/completions endpoints of the OpenAI API. Without support for /v1/responses, users cannot utilize the full range of OpenAI's latest models and features.
Desired Solution you'd like
Please add support for the /v1/responses API endpoint to ensure compatibility with all current and future OpenAI models. This enhancement will prevent the above error and allow users to leverage the complete capabilities of the OpenAI platform.
Alternatives Considered
No response
Additional Context
@jrkropp commented on GitHub (Jun 27, 2025):
I have been working on a manifold that brings support for the responses API (as a stop gap until if/when it's supported natively). You can test and monitor progress here. Supports visible reasoning summaries, optimized tool calling, web search, and much more. Working on adding deep research as well.
@tjbck commented on GitHub (Jun 27, 2025):
PLEASE check for existing discussions/issues before creating one.
@perelin commented on GitHub (Jul 22, 2025):
I have been running into the same issue with the deep-research models. @tjbck tbh: I did not find any discussion about this topic. Maybe I searched for the wrong terms?
404: This model is only supported in v1/responses and not in v1/chat/completionsis the error message that the UI shows, no discussion search results for it.@Oatu4396 commented on GitHub (Jul 23, 2025):
There's one discussion started around March suggesting adding support of responses API.
https://github.com/open-webui/open-webui/discussions/11874
@fuzedxb1 commented on GitHub (Aug 9, 2025):
Just wanted to show my support for this too, I think it's pretty essential. @jrkropp good work on what you've done to support this so far, ive used it and works perfect. Thank you.
@wizard-leoliao commented on GitHub (Oct 20, 2025):
Is this issue still unresolved? Almost all OpenAI API models except GPT‑5 are not working properly.
@Classic298 commented on GitHub (Oct 20, 2025):
Responses API is not supported and probably will not be, because it is not an open standard like the OpenAI completions API.
It's like a proprietary endpoint like Anthropic or Gemini API.
Using one of the many pipes, pipelines or other implementations out there is recommended - alternatively, the use of a middleware like LiteLLM should also do the trick
@verynewtocode commented on GitHub (Nov 5, 2025):
in your opinion, what are the chances that v1/responses is ever available