Artificial Analysis launches AA-Robotics benchmark; GPT-6 Astra tops robot arm control
ArtificialAnlys · x · 2026-11-21
Artificial Analysis has released AA-Robotics v0.1, a research-preview benchmark measuring how well frontier language models understand, plan and act in the physical world: each model directly controls a 6-axis robot arm (5 joints plus a gripper) on simple objectives, with results for the first eight tasks published today.
Key findings:
- GPT-6 Astra ranks first with a score of 69, followed by GPT-6.1 Sol, Gemini 4 Argon, Claude Opus 5.5 and Claude Fable 5.1
- GPT-6 Astra averaged 30 minutes and $16.36 per task — nearly two orders of magnitude worse than human baselines
- On the box-cleanup task, GPT-6.2 Astra used just 92 arm commands, 75% fewer than Claude Fable 5.2; across six tasks it sent 668 commands, the fewest of 11 models while scoring highest
The report concludes that robotics will likely involve strong generalist agents, though frontier models remain far too slow and expensive for production use today — something expected to change quickly.
Related event: GPT-6.2 Astra Tops New AA-Robotics Benchmark(2 posts)→
More from Embodied
- Reverse-engineering Apple Watch: a single Claude Code prompt builds a physics watch face in 30 minutes — mtizard · 2026-11-21
- Robotics Founder: Housekeeping Bots Should Target Mundane Upkeep, Not Party Tricks — Ben_Reinhardt · 2026-11-21
- GPT-6.2 Astra tops AA-Robotics v0.1 with 75% fewer arm commands than Claude — ArtificialAnlys · 2026-11-21
- Heave-Ho: decentralized framework coordinates teams of 2–10 humanoids on long-horizon tasks — yuewang314 · 2026-11-21
- USC's Heave-Ho coordinates humanoid robot teams with zero network messages, 86% success at 120x faster planning — yuewang314 · 2026-11-21
- Unitree G1 humanoid robot 'gets a job' at U-Haul in real-world demo — yuewang314 · 2026-11-21