agentjido / req_llm / 8975f67f117f9461da65570207258acc4bd2e4e9-PR-192

50%

main: 49%

Build Type

Pull #192

github

Committed by neilberkman

Commit Message

Fix: Extract cached token counts from Google API responses

Fixes #191

Google's API returns cachedContentTokenCount in usageMetadata for both
implicit (automatic) and explicit (CachedContent API) prompt caching.
The provider was not extracting this field, causing cached_tokens to
always be 0 even when caching was active.

Changes:
- Extract cachedContentTokenCount from Google responses
- Convert to OpenAI-compatible prompt_tokens_details.cached_tokens format
- Affects both google and google-vertex providers using Gemini models

Pull Request Pull Request #192: Fix: Extract cached token counts from Google API responses

Run Details

4 of 6 new or added lines in 1 file covered. (66.67%)

2 existing lines in 1 file now uncovered.

3996 of 7570 relevant lines covered (52.79%)

354.12 hits per line