OpenAI 28-Day Sprint Day 4: Sol Ultra-Fast Mode Up to 8x Speed, Codex Supports Real-Time Correction

Beating AI News Flash: OpenAI's Head of Product and Platform, Thibault Sottiaux, announced Day 4 of the 28-day product sprint. GPT-6.1 Sol Ultrafast has begun rolling out to the API, Codex, and ChatGPT Work. According to the official announcement, it possesses model capabilities approaching those of GPT-6 Astra, with generation speeds in Codex reaching up to 8 times that of Sol standard mode, and up to 6 times in the API.

Ultrafast's API pricing is $12 per million input tokens and $60 per million output tokens, both 6 times the standard tier. The API is open to all developers, but with rate limits. Codex and Work, meanwhile, are only available to $500-per-month Pro users, as well as eligible enterprise and education users.

On the same day, OpenAI also improved Codex's mid-course correction feature. On the ChatGPT desktop client, while an Agent is executing a task, users can continue to add requirements or correct direction. The model can adjust more quickly, reducing instances of continuing to work in the wrong direction.