Release v0.11.0
Changes
- fix: remove deprecated models no longer available on provider APIs (#243)
- fix: backfill resolved_provider from model when inferred (#242)
- feat: add protocol= + base_url= for compatible APIs (#241)
- chore: add co-authors for tools PR (#199) (#240)
- fix: BlockingPortal thread leak in sync stream iteration (#239)
- feat: unified tools= parameter for tool calling (#199)
- feat: add new model definitions for Google, OpenAI, and xAI (#213)
- fix: warn when unsupported parameters are silently ignored (#212)
- fix: correct model ID in README Quick Start (#211)
- refactor: make _parse_content a pure parser, add Chunk type param (#203)
- fix: enrich streaming HTTP errors with provider-specific messages (#196)
- feat: add extra_headers support to all client methods (#193)
- fix: detect and raise SSE stream error events instead of silently discarding them (#144) (#192)
- fix: empty/whitespace credentials should raise MissingCredentialsError (#190)
- refactor: make ParameterMapper generic on Content type (#189)
- fix: move _transform_output from provider _parse_content to base _predict pipeline (#187)
Installation
uv add "celeste-ai==0.11.0"