github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9659] Compatibility with new OpenAI responses API #6303

New Issue
Closed
opened 2026-04-12 17:45:23 -05:00 by GiteaMirror · 16 comments
GiteaMirror commented 2026-04-12 17:45:23 -05:00
Owner
Copy Link

Originally created by @pamelafox on GitHub (Mar 11, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9659

Originally assigned to: @drifkin, @ParthSareen on GitHub.

OpenAI just announced the responses API:
https://www.youtube.com/live/hciNKcLwSes
https://platform.openai.com/docs/guides/responses-vs-chat-completions

They are not currently deprecating chat completions, however.
I'm curious what you think about a compatibility layer with the Responses API, or if you're going to stick with chat completions compatibility only.

Originally created by @pamelafox on GitHub (Mar 11, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9659 Originally assigned to: @drifkin, @ParthSareen on GitHub. OpenAI just announced the responses API: https://www.youtube.com/live/hciNKcLwSes https://platform.openai.com/docs/guides/responses-vs-chat-completions They are not currently deprecating chat completions, however. I'm curious what you think about a compatibility layer with the Responses API, or if you're going to stick with chat completions compatibility only.
GiteaMirror added the feature request label 2026-04-12 17:45:23 -05:00
GiteaMirror commented 2026-04-12 17:45:25 -05:00
Author
Owner
Copy Link

@ParthSareen commented on GitHub (Mar 13, 2025):

Hi @pamelafox,

The team is aware of the changes and we keep a close eye on what is "standardized". Our approach always has been to provide the simplest experience for which we have our own API and SDKs.

We'll see what is worthwhile and what users need as the landscape evolves.

<!-- gh-comment-id:2722269343 --> @ParthSareen commented on GitHub (Mar 13, 2025): Hi @pamelafox, The team is aware of the changes and we keep a close eye on what is "standardized". Our approach always has been to provide the simplest experience for which we have our own API and SDKs. We'll see what is worthwhile and what users need as the landscape evolves.
GiteaMirror commented 2026-04-12 17:45:27 -05:00
Author
Owner
Copy Link

@PierreMesure commented on GitHub (Mar 17, 2025):

Great that the issue has already been posted and that you are following if the Responses API becomes an industry standard!

<!-- gh-comment-id:2728892463 --> @PierreMesure commented on GitHub (Mar 17, 2025): Great that the issue has already been posted and that you are following if the Responses API becomes an industry standard!
GiteaMirror commented 2026-04-12 17:45:28 -05:00
Author
Owner
Copy Link

@OriNachum commented on GitHub (Apr 17, 2025):

This is now blocking Codex, as it expects Responses API.

<!-- gh-comment-id:2813680394 --> @OriNachum commented on GitHub (Apr 17, 2025): This is now blocking Codex, as it expects Responses API.
GiteaMirror commented 2026-04-12 17:45:29 -05:00
Author
Owner
Copy Link

@ParthSareen commented on GitHub (Apr 17, 2025):

We're looking into it!

<!-- gh-comment-id:2813691630 --> @ParthSareen commented on GitHub (Apr 17, 2025): We're looking into it!
GiteaMirror commented 2026-04-12 17:45:31 -05:00
Author
Owner
Copy Link

@cgaspard commented on GitHub (Apr 23, 2025):

+1 we need this

<!-- gh-comment-id:2825409078 --> @cgaspard commented on GitHub (Apr 23, 2025): +1 we need this
GiteaMirror commented 2026-04-12 17:45:32 -05:00
Author
Owner
Copy Link

@OriNachum commented on GitHub (Apr 24, 2025):

@cgaspard

I built this (and verified it works) for Responses API integration.

https://github.com/teabranch/openai-responses-server

This Responses API is beyond just notation. It is a state shift.
Codex works with it (Can see here in a docker: https://github.com/OriNachum/autonomous-intelligence/tree/main/baby-tau ), so ToolUse works (Running commands)

But also in plan to add: Web Search, Code Interpreter, Computer use, Upload files and files search, MCPs, and more.

It is under MIT license.

Additionally, I plan on finding a good open source models of different sizes and train them with Tool use, so they are compatible and fit different local devices.

Does this help you?
Which features you need most?
Feel free to open issues

Edit: name changed to openai-responses-server and moved to my group's ownership for better support.
It was also deployed as a pypi package

<!-- gh-comment-id:2826340187 --> @OriNachum commented on GitHub (Apr 24, 2025): @cgaspard I built this (and verified it works) for Responses API integration. https://github.com/teabranch/openai-responses-server This Responses API is beyond just notation. It is a state shift. Codex works with it (Can see here in a docker: https://github.com/OriNachum/autonomous-intelligence/tree/main/baby-tau ), so ToolUse works (Running commands) But also in plan to add: Web Search, Code Interpreter, Computer use, Upload files and files search, MCPs, and more. It is under MIT license. Additionally, I plan on finding a good open source models of different sizes and train them with Tool use, so they are compatible and fit different local devices. Does this help you? Which features you need most? Feel free to open issues Edit: name changed to openai-responses-server and moved to my group's ownership for better support. It was also deployed as [a pypi package](https://pypi.org/project/openai-responses-server)
GiteaMirror commented 2026-04-12 17:45:32 -05:00
Author
Owner
Copy Link

@gorbypark commented on GitHub (Jun 26, 2025):

The new OpenAI Agents SDK for typescript (and I assume for Python) require this to work with ollama as well.

<!-- gh-comment-id:3009122802 --> @gorbypark commented on GitHub (Jun 26, 2025): The new OpenAI Agents SDK for typescript (and I assume for Python) require this to work with ollama as well.
GiteaMirror commented 2026-04-12 17:45:33 -05:00
Author
Owner
Copy Link

@Powerkrieger commented on GitHub (Aug 26, 2025):

Is this being worked on?

<!-- gh-comment-id:3224758916 --> @Powerkrieger commented on GitHub (Aug 26, 2025): Is this being worked on?
GiteaMirror commented 2026-04-12 17:45:34 -05:00
Author
Owner
Copy Link

@ZJCODE commented on GitHub (Aug 27, 2025):

will this be supported?

<!-- gh-comment-id:3227957864 --> @ZJCODE commented on GitHub (Aug 27, 2025): will this be supported?
GiteaMirror commented 2026-04-12 17:45:35 -05:00
Author
Owner
Copy Link

@pawelkaliniakit commented on GitHub (Sep 3, 2025):

OpenAI Agents SDK - python lib needs it.

<!-- gh-comment-id:3249051618 --> @pawelkaliniakit commented on GitHub (Sep 3, 2025): OpenAI Agents SDK - python lib needs it.
GiteaMirror commented 2026-04-12 17:45:36 -05:00
Author
Owner
Copy Link

@Powerkrieger commented on GitHub (Sep 18, 2025):

There is an issue in vllm targeting the same problem. There is some more discussion there as to whether it should actually be implemented by vllm, with reasoning against it that applies to ollama aswell.

Multiple approaches appending wrappers to provide the responses endpoint are mentioned there, including LlamaStack and Open-Responses-Server.

Since Ollama staff might decide to not implement this endpoint at all or for some time, anyone needing this might be interested in those repos or at least the discussion. Disclaimer: I have not tried any of those approaches yet.

<!-- gh-comment-id:3306473649 --> @Powerkrieger commented on GitHub (Sep 18, 2025): There is an [issue in vllm](https://github.com/vllm-project/vllm/issues/14721) targeting the same problem. There is some more discussion there as to whether it should actually be implemented by vllm, with reasoning against it that applies to ollama aswell. Multiple approaches appending wrappers to provide the responses endpoint are mentioned there, including [LlamaStack](https://github.com/llamastack/llama-stack) and [Open-Responses-Server](https://github.com/teabranch/open-responses-server). Since Ollama staff might decide to not implement this endpoint at all or for some time, anyone needing this might be interested in those repos or at least the discussion. Disclaimer: I have not tried any of those approaches yet.
GiteaMirror commented 2026-04-12 17:45:37 -05:00
Author
Owner
Copy Link

@Powerkrieger commented on GitHub (Sep 19, 2025):

Having tried (or tried to try lol) both, I can say for just using it as a wrapper, OpenResponsesServer is easier to tackle this problem, as it is straightforward in being setup. LlamaStacks documentation is lacking at the initial steps so it took me waaay too long to get anything done.

<!-- gh-comment-id:3311840316 --> @Powerkrieger commented on GitHub (Sep 19, 2025): Having tried (or tried to try lol) both, I can say for just using it as a wrapper, OpenResponsesServer is easier to tackle this problem, as it is straightforward in being setup. LlamaStacks documentation is lacking at the initial steps so it took me waaay too long to get anything done.
GiteaMirror commented 2026-04-12 17:45:38 -05:00
Author
Owner
Copy Link

@realharry commented on GitHub (Oct 19, 2025):

Considering the importance (or, dominance) of OpenAI in LLM development, Ollama API is becoming less and less useful without the full support of OpenAI APIs. I'm also one of the people who are waiting for this to be implemented in Ollama. 🙏

<!-- gh-comment-id:3419161282 --> @realharry commented on GitHub (Oct 19, 2025): Considering the importance (or, dominance) of OpenAI in LLM development, Ollama API is becoming less and less useful without the full support of OpenAI APIs. I'm also one of the people who are waiting for this to be implemented in Ollama. 🙏
GiteaMirror commented 2026-04-12 17:45:39 -05:00
Author
Owner
Copy Link

@reza8i commented on GitHub (Nov 5, 2025):

If you do want to use the Responses API you can use Hugging Face’s Responses.js proxy to convert Chat Completions to Responses API.

Alternatively, for basic examples , use their example python server

https://github.com/openai/gpt-oss?tab=readme-ov-file#responses-api

<!-- gh-comment-id:3490514612 --> @reza8i commented on GitHub (Nov 5, 2025): If you do want to use the Responses API you can use Hugging Face’s Responses.js proxy to convert Chat Completions to Responses API. Alternatively, for basic examples , use their example python server https://github.com/openai/gpt-oss?tab=readme-ov-file#responses-api
GiteaMirror commented 2026-04-12 17:45:40 -05:00
Author
Owner
Copy Link

@wigging commented on GitHub (Nov 14, 2025):

Any updates on Ollama supporting the Responses API?

<!-- gh-comment-id:3534353884 --> @wigging commented on GitHub (Nov 14, 2025): Any updates on Ollama supporting the Responses API?
GiteaMirror commented 2026-04-12 17:45:42 -05:00
Author
Owner
Copy Link

@drifkin commented on GitHub (Dec 12, 2025):

The initial implementation in #13351 is merged, will be in the next release

<!-- gh-comment-id:3644316078 --> @drifkin commented on GitHub (Dec 12, 2025): The initial implementation in #13351 is merged, will be in the next release
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label feature request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#6303
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?