Fix empty-output on hard batches: raise max_completion_tokens to 16384 (tunable) - #4
Merged
Merged
Conversation
… tunable) GPT-5.x reasoning models spend max_completion_tokens on hidden reasoning BEFORE emitting output. At 4096 a hard 50-post batch consumes the entire budget reasoning (finish_reason=length, content_len=0) -> the client raises 'empty content' -> the whole batch errors. Easy batches fit and pass, so the failure looks intermittent. Raise the default to 16384 (observed usage ~5-7k, ample headroom) and make it tunable via FINTICK_LLM_MAX_TOKENS so no model-specific ceiling needs another code change. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Reasoning models (GPT-5.x) spend
max_completion_tokenson hidden reasoning before output. At 4096 a hard 50-post batch burns the whole budget reasoning (finish_reason=length, empty content) → the client raises and the batch errors; easy batches pass, so it looks intermittent. Verified against the live failing batch: 4096→empty, 16384→full output (~5-7k used). Raises default to 16384 and addsFINTICK_LLM_MAX_TOKENSto tune per model without code changes.