GPT-6.2 Astra 在 AA-Robotics v0.1 机械臂测试中以最少指令拿到最高分
搞机器人或具身智能的朋友可以看看这个新基准,11个模型实测,GPT-6.2 Astra 用最少的机械臂指令完成任务还拿了最高分,比 Claude Fable 5.2 省了约75%的指令量。选模型接机械臂之前,这份数据挺有参考价值。
Artificial Analysis 发布 AA-Robotics v0.1 机械臂基准的首批测试结果,GPT-6.2 Astra 在 11 个受测模型中综合得分最高。完成清理箱子任务时,Astra 只用了 92 条机械臂指令,比 Claude Fable 5.2 少约 75%。
在全部六个任务中,Astra 总共发送 668 条指令,是所有受测模型中最少的,同时拿下最高分。这意味着它在保证任务成功率的前提下,把控制序列压缩到了最短,直接降低真实机器人场景的推理成本和延迟。
指令效率对机器人领域是关键指标:更少的指令意味着更少的API调用开销和更短的执行时间。对做具身智能或机器人编排的开发者来说,这份基准提供了在 OpenAI、Anthropic 等模型之间选型的直接参照。
GPT-6.2 Astra gets the job done in the fewest moves of any model we’ve tested in AA-Robotics v0.1
It cleaned up the box with 92 arm commands, ~75% fewer than Claude Fable 5.2. Across six tasks it sent 668, the fewest of 11 models, and scored the highest https://t.co/ND5u3Db7pV
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include GPT-6.2, Claude Fable 5.2, robots, embodied AI.
Source attribution: curated from Artificial Analysis (@ArtificialAnlys).