LangSmith Launches Fine-Tuning: Train Models on Your Agent Traces via Baseten
baseten · x · 2026-09-25
LangChain announced LangSmith Fine-tuning at its Interrupt 26 conference, letting developers fine-tune models on their own agent traces stored in LangSmith. The company also open-sourced smithtune, a CLI that reads LangSmith traces, converts successful runs into a supervised fine-tuning dataset, trains via the Baseten Loops Python SDK on dedicated GPUs, and deploys the evaluated checkpoint to Baseten Dedicated Inference. LangChain's applied research team already used this workflow to train the models behind LangSmith Engine; it's now open to everyone building on LangSmith.
More from coding & agent
- Open-Sourced 3D Pelican Bike Game: One Prompt, Zero Hand Edits, Prompt Included — EricBuess · 2026-09-25
- CatWalk: Claude Opus 5.5 One-Shots a Playable Game, Only Tweak Was Quieter SFX — EricBuess · 2026-09-25
- A One-Prompt Spearfishing Game on Opus 5.5, Costing 16% of a ¥3000 Plan — EricBuess · 2026-09-25
- Arkenfall: Opus 5.5 Builds a Zero-Asset Browser Open-World Game in 16 Hours — EricBuess · 2026-09-25
- One-Prompt Games: Claude Opus 5.5 Ships a Wave of Playable Browser Games — EricBuess · 2026-09-25
- Ditching Three.js for Native WebGPU: Opus 5.5 Fishing Game Hits 100FPS, Open-Sourced — EricBuess · 2026-09-25