What's Changed
- Enable the vLLM server type, with a configurable thinking token budget by @Mugga6315 in #97
- Handle for models that supply broken JSON tool-call arguments by returning a retry-response back to the model by @skye-harris in #106
New Contributors
- @Mugga6315 made their first contribution in #97
Full Changelog: 1.9.0...1.10.0