OpenAI's new Astra and Sol models reach Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, with Astra's API price cut by 20% versus GPT-6 Astra.
OpenAI on Tuesday, November 17, released GPT-6.2 Astra and GPT-6.2 Sol, two new versions of its flagship model family. Astra is pitched as the model for the hardest and longest projects, while Sol is a cheaper option for everyday coding and writing. Both are rolling out in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu subscribers, and developers can call them through the API as gpt-6.2-astra and gpt-6.2-sol. OpenAI lists Astra at $8 per million input tokens and $40 per million output tokens, which it says is 20% below GPT-6 Astra. Sol keeps its GPT-6.1 pricing of $2 for input and $10 for output.
The release comes about two months after GPT-6 Astra debuted in September and a few weeks after Sol moved to version 6.1 at DevDay. OpenAI says both new models finish demanding work with fewer output tokens, and that Astra scores 5.2 points higher than GPT-6 Astra on the Agents' Last Exam benchmark at under half the cost per task. The company also says Sol came within half a point of Astra on Cognition's FrontierCode 2.0, a test of whether code changes are good enough for maintainers to merge, at roughly one-fifth of the cost. These figures come from OpenAI and have not been independently reproduced. Both models also ship with tighter safeguards for higher-risk cybersecurity requests, which may mean more refusals for some legitimate security work.
For now the models are not being added to regular ChatGPT chat, according to an OpenAI staff reply on its developer forum, and the announcement says nothing about free users. Subscribers on the $200-a-month Pro plan get Ultrafast mode for the first time, starting with Sol, while the $500 Pro tier gets it on both models. Ultrafast generates text up to 4.5 times faster but uses up allowances more quickly, and Enterprise and Edu admins must switch it on. Users can change speed in the model picker. OpenAI says it is also working to bring both models to Daybreak, its program that gives approved security defenders fewer refusals.



