Repository navigation
fix: sync Gemini tool-use prompt tokens from Python any-llm - #29
Closed
cursor[bot] wants to merge 1 commit into
Closed
cursor[bot] wants to merge 1 commit into
cursor[bot] wants to merge 1 commit into
Conversation
Include Gemini toolUsePromptTokenCount in promptTokens so streaming and non-streaming usage categories sum to totalTokens, matching mozilla-ai/any-llm #1367. Co-authored-by: Anurag Roy <anuragroy@duck.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This was referenced Sep 10, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
mainthroughc2420fa(2 new commits since2388f59). One PR for the applicable library change.Ported
promptTokens. Gemini reports server-side tool tokens intoolUsePromptTokenCount, which is excluded frompromptTokenCountbut included intotalTokenCount. Streaming and non-streaming completions now add that field sopromptTokens + completionTokensmatchestotalTokens. Cached tokens stay a prompt-token subset; thought tokens stay completion-token details. Vertex AI inherits the same conversion.Skipped (not applicable)
tests/conftest.pyfixtures only; this repo has no matching live model map.Draft sync PR #28 is still open but was created 2026-09-04, so this is a new PR rather than a push onto that branch.
Tracked source SHA is now
c2420fac99c438518249dc56880aaee565a20efd.oxfmtandoxlintwere run after the changes.