Add bridge for /chat/completion -> /responses API #11632
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…arate folder future work to support completion_to_responses bridge allow calling codex mini via chat completions (and other endpoints)
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
krrishdholakia
changed the title
…n in separate folder" This reverts commit ff87cb8.
write it like a custom llm - requires lesser 'new' components
…and back bridge (non-streaming)
…ses bridge Allows calling codex mini via proxy
maintain <50 LOC
4 tasks
|
@krrishdholakia Thanks for taking this further! However, it seems something broke in the refactoring: The API Key is no longer used from the configuration but none is sent... (from the logs) and litellm will respond with: |
|
#11668 This fix worked for me. |
|
Hi @matthid can you share steps to repro your error? |
|
I sent a completion message to codex-mini-latest with the following proxy config: Nothing special otherwise. |
|
@matthid fixed on main 👍 |
|
@krrishdholakia Further testing revealed that the responses API is a bit more picky about the types: d2db195 Also your fix doesn't work for me since it now complains with: I actually run into the same and settled into only setting the two properties by hand (not knowing better) |
|
How are you testing this? @matthid (appreciate your help qa'ing this) |
|
@krrishdholakia Sure, I'm happy that a basic version is in main ;) Same as before via the same proxy config: The types error happens when you there is an agent message in the chat completion (I use openwebui to test this, so this happens as soon as you continue an existing chat): The hence why I used I'm not sure why we sending all of the stuff to openai, apparently only 16 fields are allowed. |
|
Thanks @matthid - i'll try to connect to OpenWebUI and test as well |
|
can i see your full config? curious if there's any other settings i should be adding |
|
Well the full setup is also quite easy: Logs from local dev env (WSL2)Request ( {
"model": "openai/codex-mini-latest",
"stream": false,
"messages": [
{
"role": "user",
"content": "What are some things to think about?"
}
]
}Logs: Deployment via dockerUsing But to be honest for debugging I just used the config, fired it up in the IDE and sent request via insomnia, pretty basic and reproduced the issue... Full litellm_configIIRC I'm using drop_params since I head issues somewhere else with openai not allowing the request, not related to this particular feature. |
|
Interesting - i'm testing claude code which uses Fixing now. |
krrishdholakia
added a commit
that referenced
this pull request
Jun 14, 2025
prevents optional params for responses api from including invalid args / litellm internal information Closes #11632 (comment)
|
@krrishdholakia cool thanks a lot! Did you also fix the |
krrishdholakia
added a commit
that referenced
this pull request
Jun 14, 2025
prevents optional params for responses api from including invalid args / litellm internal information Closes #11632 (comment)
X4tar
pushed a commit
to X4tar/litellm
that referenced
this pull request
Jun 17, 2025
* refactor(responses/): refactor to move responses_to_completion in separate folder future work to support completion_to_responses bridge allow calling codex mini via chat completions (and other endpoints) * Revert "refactor(responses/): refactor to move responses_to_completion in separate folder" This reverts commit ff87cb8. * feat: initial responses api bridge write it like a custom llm - requires lesser 'new' components * style: add __init__'s and bubble up the responses api bridge * feat(responses/transformation): working sync completion -> responses and back bridge (non-streaming) * feat(responses/): working async (non-streaming) completion <-> responses bridge Allows calling codex mini via proxy * feat(responses/): working sync + async streaming for base model response iterator * fix: reduce function size maintain <50 LOC * fix(main.py): safely handle responses api model check * fix: fix linting errors
X4tar
pushed a commit
to X4tar/litellm
that referenced
this pull request
Jun 17, 2025
prevents optional params for responses api from including invalid args / litellm internal information Closes BerriAI#11632 (comment)
|
@krrishdholakia Sorry have you seen my last comment? I'm just asking to verify if I can come back to the latest build (instead of my custom version). Thanks! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Title
Allow calling
/responsesAPI models vialitellm.completion(..)Relevant issues
allow calling codex mini via chat completions (and other endpoints)
Builds on #11440 (s/o @matthid)
Closes #9754
Closes #10925
Pre-Submission checklist
Please complete all items before asking a LiteLLM maintainer to review your PR
tests/litellm/directory, Adding at least 1 test is a hard requirement - see detailsmake test-unitType
🆕 New Feature
🐛 Bug Fix
🧹 Refactoring
📖 Documentation
🚄 Infrastructure
✅ Test
Changes