NVIDIA Opens Cosmos World Model Training Stack, Adds RL Support for Robotics
NVIDIA Developer · youtube · 2026-10-30
NVIDIA Developer published a Cosmos Labs livestream walking through building custom world models with the Cosmos Framework.
Key points:
- The framework opens up more of the training stack behind Cosmos models, with reusable skills and recipes
- Demonstrates choosing a recipe, working with compatible models, and pointing coding agents to skills for training, customization, distillation, and quantization
- New offline and online reinforcement learning support aimed at robotics development workflows
- Serving Cosmos models via Cosmos NIM while assessing quality, GPU utilization, latency, and throughput
The NVIDIA team also answers live questions on post-training and deploying Cosmos 3.
More from coding & agent
- Exo: Berkeley's self-improving agent architecture cuts costs 31-62% vs Codex — arena · 2026-10-30
- Indie dev builds Backscroll: a steerable Twitter/IG feed reranked by dirt-cheap decision models — sdand · 2026-10-30
- "It was Supabase": a viral app's creator had no idea where it was hosted — gabe_ragland · 2026-10-30
- New paper asks why teams of AI agents fail where a single agent often matches them — JoshPurtell · 2026-10-30
- Superconductor launches: multiplayer AI workspace putting teams and coding agents in one session — sergeykarayev · 2026-10-30
- YC F26 Startup Hue Launches Simulated Environments for Testing AI Agents — ycombinator · 2026-10-30