LIBERO experiments show stronger reasoning in GPT-6 variants means better robot manipulation
YuXiang_IRVL · x · 2026-09-23
USC researcher YuXiang Xu reports a striking ordering on one LIBERO task: GPT-6 Astra > GPT-6 Sol > GPT-6 Luna — better reasoning correlates with better manipulation. He argues that alongside world models, video generation and robot data, the field also needs stronger reasoning and multimodal models, linking OpenAI's reasoning models docs.
More from Embodied
- Tesla Berlin Confirms Workers Will Wear Camera Rigs to Train Optimus; Global Humanoid Shipments Top 22k in H1 — CyberRobooo · 2026-09-23
- eidon-ai releases tracker-pov robotics video dataset, 10K-100K entries — eidon-ai · 2026-09-23
- Dev turns a 90s TV into an AI TV with Raspberry Pi, mic, camera and GPT Realtime — OpenAIDevs · 2026-09-23
- Robot arm tests rank GPT-6 variants: better reasoning means better manipulation at 30x the cost — YuXiang_IRVL · 2026-09-23
- Steering VLA robot models at inference time lifts grasp rate from 14% to 74% without retraining — drmapavone · 2026-09-23
- Frontier AI models can control robots to follow harmful requests, NBC News reports — ycombinator · 2026-09-23