RAX Drives Robots via Frozen LLM Exploration at 1/20 the Cost

Researchers including Yilun Du released RAX, a robot agent system where a frozen LLM coding agent connects to robot perception and control APIs and learns purely through scaled online exploration. RAX outperforms π0.5 with up to 96.7% success at roughly 1/20 the cost of Opus.

2026-10-22 ~ 2026-10-22 · 3 related posts

• RAX drives robots with a frozen coding agent; DeepSeek version beats π0.5 at 1/20 cost — du_yilun · 2026-10-22
• RAX drives robots with a frozen LLM via pure exploration at 1/20th Opus cost — du_yilun · 2026-10-22

1 near-duplicate retellings: du_yilun