Aug 13, 2026, 10:00 AMArtificial Intelligence
OpenAI Ultrafast Runs GPT-5.6 Sol Up to 14× Faster
OpenAI previews Ultrafast, an API service tier powered by Cerebras that runs GPT-5.6 Sol up to 14× faster and reaches 750 output tokens per second in preview.
Listen to this briefingAudio briefing
Summary
OpenAI on August 13, 2026, previewed Ultrafast, a new API service tier for GPT-5.6 Sol. Powered by Cerebras, it runs the model up to 14× faster and delivers up to 750 output tokens per second.
The preview concerns OpenAI API users running GPT-5.6 Sol. The source provided only a short feed summary, leaving pricing, eligibility, rollout timing, benchmark methodology, the 14× comparison baseline and general availability unspecified.
Positives
- GPT-5.6 Sol can run up to 14× faster through Ultrafast, according to OpenAI.
- Ultrafast delivers up to 750 output tokens per second during the preview.
- Cerebras powers OpenAI’s new Ultrafast API service tier.
Risks & concerns
- The 14× claim has no disclosed comparison baseline or benchmark methodology in the available feed summary.
- Ultrafast pricing, eligibility and rollout timing were not included in the limited source.
- GPT-5.6 Sol’s service constraints and Ultrafast’s general availability remain unspecified in the supplied material.
Primary sourceOpenAI Newshttps://openai.com/index/previewing-ultrafast
Read full article