OpenAI 发布 GPT-6.1 Sol，主打以 GPT-6 Astra 五分之一的价格接近其智能水平。官方称在所有推理设置下，其错误率与 GPT-6 Astra 相差不超过 1.9%；在低推理强度下，事实错误响应占比从 GPT-6 Sol 的 11.4% 降至 7.7%。配图 DeepSWE 1.1 展示三款模型在长周期软件工程任务上的得分与单任务成本，GPT-6.1 Sol 以约 $1 成本达到约 75% 得分。原预期当天发布的 GPT-6.1 Astra 未推出。
材料汇总了 GPT-6.1 Sol 的成本定位、错误率变化和 DeepSWE 1.1 对比图，便于读者理解其与 GPT-6 Astra 的取舍。
OpenAI launched GPT-6.1 Sol.
> near GPT-6 Astra level intelligence at a fifth of the price.
> across all reasoning settings, it's error rate stays within 1.9% of GPT-6 Astra.
> It improves factual accuracy when given difficult prompts. Its largest gain over GPT-6 Sol on this front appears at low reasoning effort, where the share of responses containing a factual error drops from 11.4% to 7.7%.
Also not launching GPT-6.1 Astra, as was originally expected for today.
GPT-6.1 Sol: near-Astra intelligence for a fifth of the price. It’s the most cost-efficient model for its performance available today.在 X 查看被引用的帖子
来源：Rohan Paul · x.com