OpenAI’s new Ultrafast mode promises 14x speed for GPT 5.6 Sol, but preview access is limited

OpenAI has released a preview of “Ultrafast,” a new mode it says can make GPT 5.6 Sol — its latest and most powerful model — work at 14 times the speed of standard processing. The company described the feature as a step toward “more useful work per second” for enterprises that need fast responses without switching to a smaller model.

According to TechCrunch, which reported the launch on Thursday, Ultrafast is available only to a small group of customers in preview. OpenAI said access will expand “as capacity grows.”

What OpenAI has announced

OpenAI said in a blog post that Ultrafast is designed to accelerate GPT 5.6 Sol, its latest flagship model. The company claims the mode can deliver up to 750 output tokens per second. Tokens are the chunks of text an AI model produces, so higher tokens per second means faster generation.

The company framed this as more than a raw speed boost. “Until now, getting real-time speed typically meant choosing a smaller or more specialized model,” it said. “Ultrafast points to progress in a new direction: more useful work per second.”

The launch was reported by TechCrunch on 13 August 2026. OpenAI has not published independent benchmarks, and the figures in the current preview have not been verified externally.

What 14x speed does and does not mean

The “14x” figure is a company claim, not an independent benchmark. OpenAI says Ultrafast can work at 14x the speed of standard processing, but the launch material does not explain whether that refers to raw generation throughput, end-to-end response time, or a combination of both.

For businesses, the distinction matters. A faster model that still takes time to connect to internal systems may not feel 14x quicker in a real workflow. Enterprises should treat the number as a headline metric until OpenAI publishes more detail on how it was measured.

Where enterprises could use it

OpenAI specifically named four use cases for Ultrafast:

• Incident response, where quick answers can help teams triage problems faster.
• Customer service and support, where response time is visible to end users.
• Financial market analysis, where time-sensitive data requires fast processing.
• E-commerce, where automated assistance and product recommendations need to feel immediate.

The company did not publish a detailed pricing plan or a list of pilot customers in the announcement.

How it fits the AI speed race

OpenAI is not alone in pushing faster response times. Anthropic has launched a fast mode for Claude, though TechCrunch notes it does not reach the speed OpenAI is claiming for Ultrafast. The broader pattern is that major labs are trying to sell enterprises on speed as a core product feature.

OpenAI’s approach relies on hardware as well as software. The Ultrafast preview is powered by its partnership with chipmaker Cerebras. That means availability may depend on how much computing capacity OpenAI can secure from Cerebras as more customers ask for the mode.

Availability and the Cerebras partnership

The preview is limited to a small group of customers. OpenAI says it will expand access “as capacity grows,” but it has not given a timeline or said which geographies will get access first. The announcement did not mention India-specific availability or local pricing.

For enterprises evaluating the mode, the immediate question is whether their workloads fit the limited rollout. Since the preview is tied to Cerebras infrastructure, customers may not be able to run Ultrafast on their existing OpenAI setup until the mode becomes more widely available.

What remains unclear

Several details are still missing from the public announcement:

• Verification: The 14x speed and 750-token-per-second figures are company claims, not independent results.
• Measurement: OpenAI has not detailed how the speed gain was calculated or tested.
• Pricing: No commercial pricing for Ultrafast has been announced.
• Scale: The company has not said how many customers are in the preview, nor when wider access will begin.
• Quality: OpenAI has not said if there are trade-offs in accuracy or reliability at higher speed.

Businesses should treat the launch as a preview rather than a proven production feature. The safest next step is to ask OpenAI whether their accounts qualify for the preview, request a trial with their own workloads, and check the company’s documentation for benchmark details before committing any critical process to Ultrafast.

Item	Claim or status	Source
Mode	Ultrafast preview for GPT 5.6 Sol	OpenAI blog; TechCrunch
Claimed speed	14x faster; up to 750 output tokens per second	Company claim
Availability	Small group of customers; expansion “as capacity grows”	OpenAI
Hardware	Powered with chipmaker Cerebras	OpenAI
Enterprise use cases	Incident response, customer support, financial analysis, e-commerce	OpenAI

Source: TechCrunch, “OpenAI introduces ‘Ultrafast,’ a new mode that makes GPT 5.6 Sol work at 14x the speed” (https://techcrunch.com/2026/08/13/openai-introduces-ultrafast-a-new-mode-that-makes-gpt-5-6-sol-work-at-14x-the-speed/).