OpenAI rolls out GPT-6.1 Sol Ultrafast, an iOS-focused model claimed to run eight times faster than

the standard tier Thomas Ricouard surfaced the new mode on X, showing a live demo in which he rewrites pieces of a simple weather app in real time. The model is paired with the existing hot-reload libraries Merowing Inject and InjectionIII, so every code change appears instantly in the simulator — no fresh compile required.

The integration rewrites the usual write-build-run loop. Developers stay in the flow, getting immediate visual feedback on UI tweaks, new components, or entire features. Ricouard says the speed keeps him locked on a single task without the context-switching tax that normally piles up during front-end iteration.

Pricing sits a notch above GPT-6 Astra on the API: $12 per million input tokens and $60 per million output tokens. The model is also available in the European Union and bundled into the $500-a-month Pro plan, alongside API and Codex app access. OpenAI has not published official benchmarks beyond the "up to 8x" claim, so real-world throughput will vary with task type and context length.

No formal launch date or rate-limit details were disclosed. To try the mode, developers need an active Pro subscription or API key and must wire the injection libraries into their Xcode setup — an extra step that confines the initial audience to iOS teams already using those tools.