Repository navigation
fix: sync DeepSeek V4 Chat controls from Python any-llm - #30
Closed
cursor[bot] wants to merge 2 commits into
Closed
cursor[bot] wants to merge 2 commits into
cursor[bot] wants to merge 2 commits into
Conversation
Port mozilla-ai/any-llm #1380 so DeepSeek maps reasoning effort to the current V4 Chat thinking and effort fields, validates user_id, drops unsupported request fields, and replays assistant reasoning whenever tools are present. Co-authored-by: Anurag Roy <anuragroy@duck.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Port mozilla-ai/any-llm #1372 and #1381 so Gemini maps reasoning effort to current thinking levels and model budgets, Otari keeps structured Messages on the native endpoint, and Anthropic/Otari can stream schema-constrained Messages events. Co-authored-by: Anurag Roy <anuragroy@duck.com>
This branch was successfully deployed
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Sync with mozilla-ai/any-llm
mainthroughd137811(2 new commits sincec2420fa). One PR for the applicable library change.Ported
None/autoleave thinking and effort unset so the provider default (enabled/high) applies. Explicitnonesendsthinking: { type: "disabled" }.low/medium/high/xhigh/maxmap to DeepSeek'slow/high/maxwire efforts with thinking enabled.minimalraisesInvalidRequestError.userbecomes validateduser_id([A-Za-z0-9_-]{1,512}). Unsupported fields (frequency_penalty,logit_bias,n,parallel_tool_calls,presence_penalty,seed,service_tier) are dropped. CallerproviderOptions.thinking/user_idwin and are not mutated. Whentoolsis present, previous assistantreasoning_contentis replayed on every assistant turn (not only tool-call turns) andextra_contentstays off the wire.Skipped (not applicable)
tests/conftest.pyfixtures only; this repo has no matching live model map.Draft sync PRs #29 (created 2026-09-08) and #28 (created 2026-09-04) are still open but older than one day, so this is a new PR rather than a push onto those branches.
Tracked source SHA is now
d1378116f9fc9528892044712e0ab2bb17d0aa6e.oxfmtandoxlintwere run after the changes.