Trajectory launches post-training platform to train and continually learn trillion-parameter models
simonguozirui · x · 2026-11-03
Trajectory launched its post-training platform, billed as the easiest tool to post-train your own models and a foundation for continual learning. With a few agent prompts you get a creative suite to train, eval and deploy trillion-parameter models. Two research previews ship: Continual grading, which scales your taste so the model learns what "good" looks like beyond what you could grade by hand, and GS1, a new algorithm. cHHillee (Thinking Machines) notes Trajectory was an early Tinker customer whose experimental post-training work informed Tinker's roadmap.
Related event: Trajectory Launches Post-Training Platform for Trillion-Parameter Models(3 posts)→
More from coding & agent
- OpenClaw cut its ~$1M/month CI bill by over 95% with adaptive agentic testing — vincent_koc · 2026-11-04
- Perplexity Computer can now run Meta ads, Klaviyo email and Shopify stores end-to-end — perplexity_ai · 2026-11-04
- Filarise shows finance MCP server letting ChatGPT query your bank via Plaid — ChampionSpecialist60 · 2026-11-04
- Multi-agent messaging barely helps: 61% of agreed runs still hit merge conflicts — EliasEskin · 2026-11-04
- worldinmotion: open-source MCP server for encrypted agent-to-agent messaging across machines, with no approve tool by design — distronode · 2026-11-04
- ComfyUI App Mode studio chains FLUX to Qwen to Wan in one tab, demo run costs just $0.52 — no3us · 2026-11-04