OpenAI’s GPT-6 lineup has moved to a new version two months after launch. GPT-6.2 Astra and GPT-6.2 Sol are rolling out together in ChatGPT Work and Codex, after Astra debuted the GPT-6 generation in September and Sol was updated to GPT-6.1 at DevDay a few weeks later. Updates are arriving every few weeks now, and that pace is starting to matter as much as any single release.
The improvement OpenAI emphasizes most is efficiency. The company says both models need fewer output tokens to finish demanding tasks, so subscribers can do more before reaching their usage limits. That may matter more to everyday users than a benchmark gain. Usage limits are one of the most common frustrations with paid AI plans, and a model that does the same work with fewer tokens effectively gives subscribers more for the same price.
Astra remains OpenAI’s model for the hardest work. One example the company highlights is design matching: give GPT-6.2 Astra a screenshot of a slide and it can apply that slide’s layout, colors and even font sizes to the rest of a presentation. In Codex, OpenAI says Astra is better at changing and testing code in existing projects. It reports that early testers saw cleaner code style, more focused tests and fewer rounds of revision. Those are claims from testers OpenAI chose, not independent results.
The more interesting story may be Sol, the cheaper model aimed at everyday coding and writing. On FrontierCode 2.0, a Cognition benchmark that checks whether changes to real codebases are good enough for maintainers to merge, OpenAI says Sol came within half a point of Astra at roughly one-fifth of the cost, with both running at maximum reasoning effort. If independent testing confirms that, many developers will have little reason to pay for the top model.
Both models are available now on Plus, Pro, Enterprise and Edu plans. OpenAI is also widening access to Ultrafast mode, which it says produces tokens up to 4.5 times faster than the standard rate but uses up allowance more quickly. On the $200-a-month Pro plan, Ultrafast now works with Sol for the first time. The $500 Pro tier gets it on both models, as do eligible Enterprise and Edu workspaces once an admin turns it on.
Developers can use both models through the API today. Astra’s standard input and output prices are 20 percent lower than GPT-6 Astra, while Sol keeps its GPT-6.1 pricing. Cheaper top-end pricing and a strong lower-cost model point to competition moving from benchmark scores to cost and output per dollar.
