Next-Gen AI Speed
Cerebras and OpenAI Deliver Ultrafast GPT-5.6 Sol, Boosting AI Performance
Key Points
Ultrafast waitlist criteria: Businesses can apply to join Ultrafast via a waitlist by sharing workload, latency requirements, expected usage, and other details.
Expanded set of high-speed workflows supported by Ultrafast, including voice, customer support, commerce, developer agents, financial research, and security response.
Andrew Feldman, CEO and co-founder of Cerebras, described Ultrafast as 'GPT-5.6 Sol on Ultrafast is proof that speed and intelligence are no longer mutually exclusive.'
Ultrafast is claimed to be faster than Claude Opus 4.8, with about 5x faster in Fast mode and 11x faster in another measure according to coverage comparing to Anthropic/Claude.
OpenAI and Cerebras are launching a new service tier called Ultrafast, which runs GPT-5.6 Sol at up to 750 output tokens per second — up to 14 times faster than standard processing, according to Unite.ai. The tier aims to bring frontier-level AI speed to time-sensitive tasks without sacrificing model quality.
Access is currently limited. OpenAI is opening a waitlist for select businesses, asking applicants to share their workload details, latency needs, and expected usage before granting entry, Quiver Quant reported.
750 Tokens Per Second Sets a New Speed Bar
Cerebras is the hardware engine behind Ultrafast. The company's inference technology runs GPT-5.6 Sol at speeds that Unite.ai says crush the competition. Ultrafast is roughly 5 times faster than Anthropic's Claude Opus 4.8 in Fast mode and about 11 times faster by another measure.
Cerebras CEO Andrew Feldman put it plainly: "GPT-5.6 Sol on Ultrafast is proof that speed and intelligence are no longer mutually exclusive." That claim — that a model can be both very fast and very smart — has long been a sticking point in AI deployment.
Why Speed Matters for Real-World AI Workflows
Low latency — meaning less waiting time between a request and a response — is critical for many business uses. OpenAI and Cerebras say Ultrafast opens the door for live voice calls, customer support chats, financial research, and security incident response, according to Unite.ai.
Developer agents and e-commerce tools are also on the supported workflow list. When a model takes too long to respond, companies often have to choose a smaller, less capable model. Ultrafast is designed to remove that tradeoff.
Cerebras and OpenAI Bet on Specialized Hardware
The Ultrafast launch is part of a broader trend: pairing powerful AI models with purpose-built chips. Cerebras makes wafer-scale processors designed specifically for fast AI inference. Standard cloud chips were not built with this kind of speed in mind, Neowin noted.
Cerebras (NASDAQ: CBRS) announced the partnership publicly, according to Seeking Alpha. The deal gives OpenAI a speed edge while giving Cerebras a high-profile showcase for its hardware at the frontier of AI performance.
Limited Preview Now, Broader Rollout Planned Later
OpenAI is not opening Ultrafast to everyone at once. The company says it wants to learn from early users before expanding access. Businesses must apply through a waitlist and provide details about their use case, expected volume, and latency requirements, per Quiver Quant.
The cautious rollout mirrors how OpenAI has handled other major launches. By starting small, the company can catch problems early and tune the service before scaling. No public timeline for wider access has been announced.
Sentiment
Mostly PositivePublishers
18
Articles
22
Reach
40
