Unverified50% confidenceFactExact time
Ultra Fast was originally planned to run on Cerebras chips, but because Cerebras capacity was bought out by Jane Street, OpenAI could only sustain the speed by over-provisioning with NVIDIA chips.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/7/2026
First Seen
Valid until: 1/5/2027
Sources
Related Entities
Related Claims
UnverifiedThe Ultrafast inference service is powered by compute from chip manufacturer Cerebras.69% similarUnverified8月13日OpenAI预览了Ultrafast，一个全新API层级，运行旗舰模型速度最高可达常规14倍，约每秒输出750个Token，运行在Cerebras硬件上67% similarUnverifiedCerebras is a chip company focused on AI inference acceleration, and its flagship WSE (Wafer Scale Engine) is currently the world's largest single-die chip.67% similarUnverifiedOpenAI has hinted that the Sol model can achieve a significant boost in inference speed when paired with Cerebras chips.63% similarUnverifiedCerebras是一家专注于AI推理加速的芯片公司，其旗舰产品WSE针对传统GPU的内存墙痛点62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/985909API
curl https://kongchang.com/api/v1/knowledge/claims/985909MCP
get_claim(id=985909)