PartnershipAug 13, 2026, 01:01 PM
Cerebras Powers OpenAI GPT-5.6 Sol Ultrafast Mode, 14x Faster
AI Summary
Cerebras Systems Inc. announced a significant partnership with OpenAI, powering the new "Ultrafast mode" for the GPT-5.6 Sol API. This new service tier enables GPT-5.6 Sol to run up to 14 times faster than standard processing, achieving up to 750 output tokens per second, and is significantly faster than competing models. The collaboration aims to deliver frontier AI intelligence with unprecedented speed, addressing a key constraint in AI adoption. This speed is attributed to Cerebras' Wafer-Scale Engine architecture, which eliminates memory-bandwidth bottlenecks.
Key Highlights
- Cerebras powers OpenAI's new Ultrafast mode for GPT-5.6 Sol API.
- Ultrafast mode runs GPT-5.6 Sol up to 14x faster than Standard processing.
- Achieves up to 750 output tokens per second for GPT-5.6 Sol.
- 5x faster than Claude Opus 4.8 (Fast mode) and 11x faster than Claude Fable 5.
- Completed "Humanity’s Last Exam" benchmark in just over 11 hours, 7x faster than Claude Fable 5.
- Delivered a 5.6x end-to-end speedup on the GDP-Val benchmark.
- Speed is enabled by Cerebras' Wafer-Scale Engine architecture with 44 GB on-chip SRAM.
Price Impact
More from CBRS