fix: Correct cache token handling for Anthropic API semantics #316
+162
−43
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
Anthropic's input_tokens only contains NEW tokens (excludes cached), while OpenAI's prompt_tokens includes cached tokens. The old code incorrectly clamped cache values to input_tokens, causing cached_tokens to be truncated (e.g., 5484 to 12). The fix detects API format based on field presence: - OpenAI format (has prompt_tokens_details.cached_tokens): clamp cached <= input - Anthropic format (has cache_read_input_tokens): no clamping (values are additive) Co-authored-by: Scott Helvick <4016577+shelvick@users.noreply.github.com> Amp-Thread-ID: https://ampcode.com/threads/T-019bc40f-f436-712b-89dd-e28a616bc850 Co-authored-by: Amp <amp@ampcode.com>
mikehostetler
force-pushed
the
fix/clamp-tokens-anthropic
branch
from
be5533e to
49151e2
Compare
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Outdated suggestions cannot be applied.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.
Summary
input_tokensonly contains NEW tokens (excludes cached), while OpenAI'sprompt_tokensincludes cached tokensinput_tokens, causingcached_tokens: 5484to becomecached_tokens: 12Solution
Detects API format based on field presence (not provider names):
prompt_tokens_details.cached_tokens): input INCLUDES cached, so clampcached <= inputcache_read_input_tokens): input EXCLUDES cached, no clamping needed (values are additive)Test plan