fix(llma): Langchain cache token double subtraction for non-Anthropic providers #369
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
PR #346 introduced cache token subtraction for all providers, causing double subtraction for OpenAI/OpenRouter and resulting in negative costs. The plugin-server only subtracts cache tokens for Anthropic providers (exact match on provider name or substring match on model name). This fix aligns the Python SDK with that behavior. Changes: - Only subtract cache tokens when provider="anthropic" OR model contains "anthropic" - Passes provider and model metadata to usage parsing functions - Updates tests to reflect correct behavior (no subtraction for OpenAI) - Adds test for Anthropic provider subtraction Fixes negative cost calculations for users on OpenAI/OpenRouter with cached tokens.
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
andrewm4894
approved these changes
Nov 11, 2025
carlos-marchal-ph
deleted the
fix(llma)/cache-token-double-subtraction
branch
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Outdated suggestions cannot be applied.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.
Langchain internally transforms Anthropic token counts to an OpenAI like format (input tokens include in cached tokens). This was causing errors in our logic since we already undo this transformation in the plugin server. A fix for this was merged in #346.
Unfortunately the fix was too eager and it also started applying to other, non-Anthropic providers, like OpenAI itself. This causes incorrect costs (even negative ones) for generations with cache usage.