Breaking changes
All releases prior to V1 can contain breaking changes according to semvar, but this release contains more than usual:
- The format of messages has changed significantly since the last release, see #232 and #259
- more messages/parts are added that before as a result of adding
ToolReturnPartfor each tool call, see #184 and #274 - given how much we've changed since the last release, other things have probably changed along the way
What's Changed
- Minor docs updates by @sydney-runkle in #159
- use sqlite for chat_app example by @samuelcolvin in #107
- Rename tool_id to tool_call_id by @dmontagu in #206
- upate docs to include remote server example by @cal859 in #210
- Add new llama-3.3-70b-versatile to GroqModelName by @kenjihikmatullah in #202
- Documentation: minor spelling and phrasing fixes by @carlsonp in #212
- Adding a minimal contributing guide by @sydney-runkle in #213
- Add note about jupyter workaround for conflicting event loops by @sydney-runkle in #214
- Standardize "py" -> "python" in code blocks with pre-commit by @sydney-runkle in #217
- Add Anthropic (non-streaming) Support by @sydney-runkle in #193
- fix double anthropic reference by @montasaurus in #222
- Add support for new Gemini model 'gemini-2.0-flash-exp' by @Etelis in #223
- Move description onto tool when appropriate by @samuelcolvin in #228
- fix formatting of
tests/test_tools.pyby @samuelcolvin in #229 - Move function tools docs to a new page by @samuelcolvin in #231
- added makefile help target by @janas-adam in #226
- [docs] update contributing guide to indicate minimum
uvversion by @zzstoatzz in #236 - Support tool calling when a structured result is provided by @jlowin in #184
- Basic
ModelSettingslogic by @sydney-runkle in #227 - Unify model responses by @dmontagu in #232
- Disable pyright reportUnnecessaryIsInstance by @dmontagu in #244
- pin ollama action by @samuelcolvin in #246
- Mistral Support by @YanSte in #173
- feat: add optional base_url kwarg to OpenAIModel by @sambarnes in #243
- Update install.md for Ollama by @Navanit-git in #225
- live tests for mistral by @samuelcolvin in #249
- Change: Internal Discriminator Modification by @sydney-runkle in #247
- add model documentation by @samuelcolvin in #250
- Formatting nitpicks in
messages.pyby @sydney-runkle in #255 - Get chat app working with new messages format by @sydney-runkle in #251
- Use
defer_buildwith type adapters, not custom_LazyTypeAdapterby @sydney-runkle in #253 - Use
isinstancechecks for message kinds by @sydney-runkle in #252 - add
messagestoRunContextby @samuelcolvin in #257 - Reformat messages so messages become simple
list[ModelRequest | ModelResponse]by @samuelcolvin in #259 - Fix typo in agent documentation example by @Viicos in #271
- Enhance Mistral: Improved Test Coverage, Code Optimization, and Bug Fixes by @YanSte in #263
- Streamed response messages by @samuelcolvin in #274
- Fix doc error in ollama.md by @asmith26 in #279
- Alternative intro by @samuelcolvin in #270
- uprev to v0.0.13 by @samuelcolvin in #280
New Contributors
- @sydney-runkle made their first contribution in #159
- @kenjihikmatullah made their first contribution in #202
- @carlsonp made their first contribution in #212
- @montasaurus made their first contribution in #222
- @Etelis made their first contribution in #223
- @janas-adam made their first contribution in #226
- @YanSte made their first contribution in #173
- @sambarnes made their first contribution in #243
- @Navanit-git made their first contribution in #225
- @Viicos made their first contribution in #271
- @asmith26 made their first contribution in #279
Full Changelog: v0.0.12...v0.0.13