AI Models
13 August 2026
OpenAI Accelerates GPT-5.6 Sol to 750 Tokens per Second
N
New-ZZZ desk
X @OpenAI · 2 days ago
OpenAI has unveiled a preview of Ultrafast mode for GPT-5.6 Sol. It speeds up response generation by up to 14 times and can deliver up to 750 tokens per second. The mode runs on Cerebras’ computing platform.
Ultrafast will initially be available through the OpenAI API to a small group of customers. The company plans to gradually expand access for enterprises as computing capacity increases.
OpenAI will explore which products and workflows benefit businesses most from higher speeds. Companies that need fast access to the model’s capabilities can request to be notified when it launches.
Why it matters
- —The mode could accelerate AI products and workflows where response time is critical.
- —The partnership with Cerebras demonstrates an alternative approach to accelerating frontier models.
- —The limited rollout will allow OpenAI to assess the practical value of high-speed generation for businesses.
Key facts
- Ultrafast accelerates GPT-5.6 Sol by up to 14×.
- The claimed speed reaches 750 tokens per second.
- A limited group of customers will receive early access through the OpenAI API.
- Enterprise access will expand as compute capacity grows.
The full text is in the original source. Here we provide a brief summary and key facts.