LangChain launches smithtune in public beta: fine-tune models on your LangSmith agent traces
LangChain · x · 2026-09-25
LangChain announced LangSmith Fine-Tuning and the smithtune CLI in public beta. The idea: LangSmith processes hundreds of millions of agent traces daily, and smithtune turns those trajectories into high-quality SFT data via an end-to-end pipeline covering curation, preparation, training and deployment — yielding task-specialized models with lower cost and latency. Training APIs from Baseten and Fireworks are increasingly accessible, but sourcing quality data remains the bottleneck, which is the gap smithtune fills. The team frames post-training on your own agent data as a key way for teams to own their intelligence as open frontier models advance.
More from coding & agent
- Four steps to stop Opus 5.5 from wasting tokens: effort, prompt audit, CLAUDE.md rules — dr_cintas · 2026-09-25
- Nat Friedman fires back at Verge: those 'leaked secrets' are just files on the user's VM, intended behavior — dtrinh · 2026-09-25
- ServingStudio: simulate LLM serving configs before burning expensive GPU time — bariskasikci · 2026-09-25
- issue-graph prioritizes your backlog to coordinate coding agents, now on GitHub — Chamoen · 2026-09-25
- Claude now spots bugs, gathers logs, and asks users to submit packaged reports — ColleenMBrady · 2026-09-25
- SkillRL (NeurIPS 2026): 7B model beats GPT-4o by 41% via recursive skill evolution — cihangxie · 2026-09-25