Unverified50% confidenceReleaseExact time
8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
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/18/2026
First Seen
Valid until: 11/16/2026
Sources
Related Claims
UnverifiedThe flagship model Sol runs on Cerebras wafer-scale chips with an inference speed of 750 tokens/second, approximately 15 times that of the previous generation.69% similarUnverifiedFastAPI的单机并发吞吐量可达传统同步框架（如Flask）的数倍至十倍以上68% similarUnverifiedSol will go live on the Cerebras platform with an inference speed of approximately 750 tokens per second, roughly 3 to 5 times that of traditional GPU clusters.65% similarUnverifiedThis GDScript+Vulkan solution runs at approximately 1/10 the speed of the llama.cpp+CUDA solution.62% similarUnverified开源物理模拟库RocketSim能以约100倍实时速度在CPU上复现《火箭联盟》的物理行为62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/771251API
curl https://kongchang.com/api/v1/knowledge/claims/771251MCP
get_claim(id=771251)