mirror of
https://github.com/fabro-sh/fabro.git
synced 2026-09-30 01:53:45 +00:00
Extend the OpenAI-compatible adapter's ApiUsage parsing to read prompt_tokens_details.cached_tokens and cache_write_tokens from response and streaming chunk payloads. Both fields are optional, so this is fully backwards-compatible with providers that don't emit them. Cached and freshly-read tokens are priced differently, so surfacing both keeps catalog cost math accurate. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| crates | ||
| packages/fabro-api-client | ||