GPT-6-Astra 机器人基准接近饱和，但物理验证与纠错仍是短板
有人把 GPT-6-Astra 扔进 12 个机器人基准实测，分数很高但一个回合烧 900 万 token 跑两小时，落地还早。
一篇 arXiv 论文（编号 2610.22471）测试了 GPT-6-Astra 在 12 个机器人基准上的零权重更新表现。该模型在 LIBERO 套件接近饱和，RoboDojo 达到 81.9%，DexJoCo 达到 87.3%，恢复错误时所需动作步数最高是常规的 10 倍。但单个 LIBERO-Ultra-Long 回合消耗约 900 万 token、耗时近 2 小时，且在真实机器人上无法用更多步骤换取任务成功。论文结论是它目前只适合作为研究用的机器人控制器，落地部署受限于算力成本和物理纠错能力。
New paper finds that GPT-6-Astra nearly saturates easy robot benchmarks but still fails to verify physical success and recover from irreversible mistakes. And at a cost too high for real use.
Robot manipulation and navigation have been dominated by task-specific models. The question was whether 1 general model could handle them all without weight updates.
The authors tested GPT-6-Astra on 12 robot benchmarks. It studied training demos into a frozen knowledge package and got up to 10 times the usual action steps to recover.
It neared saturation on the LIBERO suite and reached 81.9% on RoboDojo and 87.3% on DexJoCo. But a single LIBERO-Ultra-Long episode used about 9M tokens and nearly 2 hours, and on real robots it could not trade more steps for success.
Treat general models as capable robot controllers for research, but expect task-specific demos, heavy compute, and physical limits to block deployment for now.
– arxiv. org/abs/2610.22471
Title: "Has GPT-6-Astra Achieved Embodied Intelligence?"
