gpt-6VIDEO
OpenVLHarness:GPT-6 Astra 视觉推理平均提升 9.9 分
新框架 OpenVLHarness 把任意多模态模型变成基于证据的视觉推理器,GPT-6 Astra 在 22 个数据集上平均涨 9.9 分,比 Codex 等 VL harness 高 7.7–8.5 分且成本不超过 37%。
Yu (Bryan) Zhou
@yu_bryan_zhou9 天前收录原推发布于 2026.10.21
Yu (Bryan) Zhou @yu_bryan_zhou原推发布于 2026.10.21
🚀 Introducing OpenVLHarness: Turn any MLLM into an evidence-based visual reasoner that thinks and codes in visual space. 📈 GPT-6 Astra gains +9.9 points on average over 22 datasets. ⚡ Beats Codex and other VL harnesses by 7.7–8.5 points, costing ≤37% as much.
x.com/yu_bryan_zhou/status/2112594601752277444采集于 2026.10.21 10:06