Sherwin Wu· @sherwinwu · X·· 2026-08-14AI 评分43
AI 导读
Cerebras 预览 GPT-5.6 Sol Ultrafast 模式，生成速度最高 750 tokens/秒，比标准处理快 14 倍。该模式跑完 Humanity's Last Exam 仅用 11 小时 11 分，比 Claude Fable 5 快近 7 倍且准确率相当。用户称瓶颈已从模型转向工具本身。
正文 · 原文
GPT-5.6 Sol Ultrafast is here!!
750 tokens per second honestly feels instantaneous for most workloads. The bottleneck now moves to the tools themselves.
Previewing Ultrafast mode for @OpenAI's GPT 5.6 Sol, powered by Cerebras. GPT-5.6 Sol Ultrafast generates responses at up to 750 tokens per second. That's the full, GPT-5.6-Sol model -- up to 14× faster than the same model on Standard processing. It speedran Humanity’s Last Exam in 11h 11m, nearly 7× faster than Claude Fable 5 with comparable accuracy.在 X 查看被引用的帖子
来源：Sherwin Wu · x.com