What's New in V1.0.3
A big release with over 200 commits spanning many features and models.
- Improved Assistant: Kiln Assistant can now call APIs with streaming status, like evals
- Input Transforms: enables "feature engineering" in Kiln. Convert your model inputs using Jinja templates, before they reach the model. Can strip unnecessary content, truncate long fields, check equality, and much more.
- New Models: GLM 5.2 , Opus 4.8, MiniMax M3, Gemini 3.5 Flash, Gemini 3.1 Flash Lite, Nemotron 3 Ultra, Step 3.7 Flash, Qwen 3.7 Plus, Minimax M3, MiMo-V2.5, thinking levels updated for Anthropic/Deekseek/Gemma model, Fable 5 (for those who made requests in those magical 4 days), "Gemini Enterprise" provider
- And More: visual design updates, UX polish, bug fixes, updated fine-tune model list, remove OpenAI fine-tuning (OpenAI deprecated), audio/video extraction on OpenRouter, and more
CI Build Source for this release Mac and Linux: /Kiln-AI/Kiln/actions/runs/28038920795
CI Build Source for this release Windows: /Kiln-AI/Kiln/actions/runs/28038920777
Full Changelog: v1.0.2...v1.0.3
New Contributors
- @ianjamesburke made their first contribution in #1456
Contributors
ianjamesburke