Back
OpenAINovember 17, 20263 sources

OpenAI launches GPT-6.2 Astra and Sol with a 20% Astra price cut; power users pan Ultrafast pricing

AI Analysis

GPT-6.2 ships as a pair. Astra is the frontier model for complex projects. It has a 1,050,000-token context window, standard input and output rates 20% below GPT-6 Astra, improved multistep-work benchmark scores, and a score of 62 on the Artificial Analysis Intelligence Index. That places it second behind Anthropic's Fable 5.2, according to Trending Topics. Sol is the efficient tier: it keeps GPT-6.1 Sol pricing while reportedly matching the previous flagship at about 15% of the cost. Both are rolling out in ChatGPT Work and Codex.

The API pricing has several tiers. Batch processing is half price, and Fast costs 2x standard. Ultrafast starts at 3x and rises to 6x on March 1, 2027. OpenAI also introduced Astra for Law, which adds an expanded legal search index and ranked first on Artificial Analysis's Harvey LAB-AA benchmark, earning full credit on 20.8% of evaluations. Separately, it made a free CSAM image classifier available via API to vetted organizations.

Reception has been split. The launch drew 1,107 points and 707 comments on Hacker News. VentureBeat quoted power users calling the models 'essentially unusable,' with much of the anger aimed at the Ultrafast price escalation. Wharton's Ethan Mollick urged restraint: '6.2 is not terrific but that doesn't tell us much,' noting that Anthropic and Google both had weak stretches before recovering.

Watch whether OpenAI revisits the 6x Ultrafast schedule before March. AWS made both models available on Bedrock on launch day, which widens enterprise reach and also exposes pricing to more direct comparison.

Sources
AI Briefing
·Vendors·Curated by AI agents · Updated daily · 2026
Built by Koby Almog