Release v0.2.7
Changes
- feat: add video-generation package with OpenAI, Google, and ByteDance providers (#43)
- refactor: add Chunk type parameter to Stream class for type-safe streaming (#42)
- refactor: remove redundant StrEnum type annotations from ParameterMapper (#41)
- refactor: migrate Anthropic to native structured outputs API and remove raw_event metadata (#40)
- chore: bump version to 0.2.6 (#38)
- refactor: improve error messages and metadata handling (#39)
- feat: add GPT-5.1 model support (#37)
- refactor: restructure finish_reason handling and add verbosity parameter (#36)
- feat: add ELEVENLABS provider and bump version to 0.2.5 (#35)
- chore: bump version to 0.2.4 and add speech-generation optional dependency (#34)
- chore: bump version to 0.2.2 and add image-intelligence optional dependency (#32)
- refactor: improve code quality and standardize MIME types (#31)
- Add GPT-5 mini and nano model to OpenAI provider (#29)
- fix: add BYTEDANCE_API_KEY to integration test workflow (#28)
- feat: add image-generation package with OpenAI, Google, and ByteDance providers (#27)
- fix: preserve finish_reason in streaming final chunk (#26)
- Replace generic errors with custom exceptions in text-generation (#25)
- implement custom exceptions for Celeste and add unit tests (#23)
- refactor: standardize capability naming to kebab-case (#24)
Installation
uv add "celeste-ai==0.2.7"
