Rohan Paul· @rohanpaul_ai · X·· 2026-07-09AI 评分51
AI 导读
GPT 5.6 Sol 在 Cerebras 专用推理硬件上获得 10 倍速度提升。Sam Altman 在六月透露，GPT 5.6 Sol 推理速度可达 750 tokens/sec。当前 GPT-5.5 优先和规模层服务广告声称 99% 大于 50 tokens/sec，因此 Sol 在 Cerebras 上可达该速率的 10–15 倍。Cerebras 的晶圆级芯片设计大幅降低了内存与网络延迟，从而实现超高速推理。
正文 · 原文
So, GPT 5.6 on Cerebrus is getting 10x speed.
This will be incredible.
In June, Sam Altman said they are getting 750 tokens/sec for GPT 5.6 Sol.
The current GPT-5.5 priority and scale-tier service advertises 99% >50 tokens/sec, so Sol on Cerebras can be up to 10-15x that rate.
This huge number is coming from the specialized inference hardware: Sol is being served on Cerebras, whose wafer-scale chip is designed to move model data with far less memory and networking delay than a normal multi-GPU setup.
@devberkayy 5.6 on Cerebras is insane. 10x speed.在 X 查看被引用的帖子
来源：Rohan Paul · x.com