Cerebras Clocks Humanity's Last Exam at 11 Hours 11 Minutes for 2,500 Questions — Against 78+ Hours for Claude Fable 5

Cerebras Blog (corroborated by unite.ai and 9to5Mac)·medium signalXBlueskyLinkedInCopy link

The most concrete number in the Cerebras/OpenAI Ultrafast writeup is a wall-clock eval comparison: GPT-5.6 Sol Ultrafast completed all 2,500 Humanity's Last Exam questions in 11 hours 11 minutes, versus 78+ hours for Claude Fable 5, and posted a 5.6x end-to-end speedup on GDP-Val with no measured degradation. Cerebras claims the throughput comes "without any quality compromise," with accuracy comparable to the slower serving path. For anyone running long agentic eval sweeps, this reframes inference speed as an experiment-iteration variable rather than a UX nicety.