Nenya 0.9.0
·
61 commits
to main
since this release
Changelog
- 97f0281 Merge phase-010: cache token accounting
- 38ea444 Merge phase-011: cache usage breakdown
- 1384dce Merge phase-012: cache TTL and per-breakpoint defaults
- 9e8c34c Merge phase-013: automatic caching mode
- 1962e06 Merge phase-014: OpenAI GPT-5.6+ cache breakpoints
- b11a30d Merge phase-015: cache_salt for vLLM-compatible tenant isolation
- 3655ce9 feat(cache): add OpenAI GPT-5.6+ explicit cache breakpoints
- c779ec9 feat(cache): add automatic caching mode with top-level cache_control
- 6032ee3 feat(cache): add cache_salt for vLLM-compatible tenant isolation
- b113924 feat(cache): parse cache_creation breakdown and iterations array
- e9c1c5e feat(cache): support 1-hour TTL and per-breakpoint TTL
- 260580d fix(cache): correct Anthropic token counting and refactor UsageCallback to UsageData
- ca325ba fix(stream): address code-reviewer findings from Phase 010 review
- 865eba8 refactor(stream): extract usage parsing helpers to reduce nesting