GPT-6 Astra 在宜家家具组装错误识别基准上达 80%
原标题：OpenAI's GPT-6 Astra can now tell you exactly where you screwed up your IKEA shelf
AI 摘要
Epoch AI 推出家具组装基准 FAB，用三张宜家家具组装照片（含故意错误）测试模型识别装配错误的能力。2025 年 11 月最佳模型 Claude Opus 4.5 得分 28%，十个月后 OpenAI 的 GPT-6 Astra 达到 80%（每张照片耗时三分钟），Claude Fable 5.1 为 70%，Claude Opus 5 为 61%，中国开源权重模型如 Kimi K3 落后领先者至少七个月。研究者称该技术目前仍太慢，无法实时辅助组装，但未来或可用于汽车维修或家电修理，且 Astra 在视觉机器人任务上也表现出色。
正文节选
OpenAI's GPT-6 Astra can now tell you exactly where you screwed up your IKEA shelf AI can now spot when you've built your IKEA furniture wrong. Epoch AI's Furniture Assembly Benchmark (FAB) photographs three IKEA pieces during assembly with deliberate errors. Models compare photos against instructions, identify mistakes, and describe what went wrong. In November 2025, the best model (Claude Opus 4.5) scored 28%. Ten months later, OpenAI's GPT-6 Astra hits 80% at three minutes per photo. Claude F