GPT-6 tested on LIBERO robot task: turns on stove, fails to grasp moka pot
YuXiang_IRVL · x · 2026-09-23
Researchers benchmarked three models on the same LIBERO robot arm task with identical prompts. GPT-6 Sol successfully turned on the stove but failed to grasp the moka pot — better than Luna, yet still unable to complete the task. Cost details: 66 API calls per episode, $2.69 total ($0.041 per call).
More from Embodied
- OpenRoboto Shift launches: decentralized egocentric video data network for robot brains — markjeffrey · 2026-09-23
- OpenRoboto launches decentralized rival to Figure's robot data network on Bittensor — markjeffrey · 2026-09-23
- Typesafe's decision model Jev drives a Go2 robot inside NVIDIA Isaac Sim via OM1 — paigeinsf · 2026-09-23
- Tesla pitches Optimus as the first general-purpose humanoid robot — XFreeze · 2026-09-23
- XPENG's IRON robot demos full-duplex speech with 9-mic array and lip reading — ChrisGPT · 2026-09-23
- Uber riders can now hail Waymo robotaxis on Austin freeways — ATTlKA · 2026-09-23