claudeVIDEO

Blender 任务跑分：GPT-6 Astra 0.686 对 Opus 5.5 0.642

20 个 Blender 任务的平均奖励对比，Astra 0.686 略胜 Opus 5.5 的 0.642，并指出 Opus 在走钢丝任务上让牛仔双脚不动只拿 0.400。

Refresh

@refresh_dev2026.10.01收录原推发布于 2026.10.01

Refresh @refresh_dev原推发布于 2026.10.01

GPT-6 Astra averaged 0.686 reward across 20 Blender tasks, Claude Opus 5.5 0.642. Astra hit 0.90+ on 7/20, Opus on 3/20. Where they fall short: on this tightrope task, Opus 5.5 kept the cowboy's feet planted instead of stepping and scored 0.400.

x.com/refresh_dev/status/2105376464648773946采集于 2026.10.01 10:16