v0.11.1

Hydaelyn v0.11.1

Status: Released

Released on 2026-07-18.

Cached input token usage

Provider usage now reports cached input tokens separately so applications can
measure prompt-cache effectiveness without changing total input accounting:

• provider.Usage.CachedInputTokens exposes the provider-reported cached
  subset of input tokens;
• provider.Usage.Add accumulates cached input tokens across streamed usage;
• OpenAI Chat Completions reads
  prompt_tokens_details.cached_tokens;
• OpenAI Responses reads input_tokens_details.cached_tokens.

The change is backward-compatible: the new field is additive and remains zero
when a provider does not report cache usage.

Upgrade

go get github.com/Viking602/go-hydaelyn@v0.11.1
go install github.com/Viking602/go-hydaelyn/cmd/hydaelyn@v0.11.1

What's Changed

• feat(provider): report cached input tokens by @Viking602 in #40
• chore(release): prepare v0.11.1 by @Viking602 in #41

Full Changelog: v0.11.0...v0.11.1