OpenAI releases GPT-6.2 Astra and Sol models, cuts Astra price by 20%
OpenAI has introduced two new models: GPT-6.2 Astra for complex projects and GPT-6.2 Sol for everyday work. According to the company, using Astra via the API has become 20% cheaper, while Sol pricing remains the same.
Per million tokens in the API, Astra costs $8 for input and $40 for output. Cached reads cost $0.40, which is 60% less than before. Sol prices remain unchanged at $2 and $10. According to OpenAI documentation, both models accept a context of up to 1.05 million tokens and can generate responses up to 128,000 tokens long. The models are already available in ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu subscriptions (according to iXBT.com, also Business). In the API, they are named gpt-6.2-astra and gpt-6.2-sol.
According to OpenAI's own benchmarks, Astra scored 5.2 points higher than GPT-6 Astra on the Agents' Last Exam test. In GDPval v2, the gain was 6.2 points, while token consumption per response decreased by 22%. In FrontierCode 2.0, Sol trailed Astra by about 0.5 points, while the task cost roughly five times less. In the independent Artificial Analysis ranking, Astra scored 62 points, ranking second after Anthropic's Claude Fable 5.2 (65 points). Sol scored 57 points. One benchmark task cost Astra around $4, compared to about $27 for Fable 5.2. In programming, Astra remained at the level of GPT-6 Astra.
On OpenAI's internal scale, Astra is the company's first released model to be assigned a high risk level in the AI self-improvement category. In cybersecurity, both models were assigned a critical level. Because of this, the models respond more strictly to hacking-related requests. Relaxed restrictions are provided only for verified cybersecurity specialists under the Daybreak program.
What it means
The models are already available in ChatGPT Work, Codex, and the API. In the $500 Pro plan, the Ultrafast accelerated response mode works for both models. In the $200 Pro plan, this mode has appeared for the first time, but so far only for Sol. If more than 272,000 tokens are submitted as input, the price of input tokens and cache reads doubles, while output token pricing increases by 1.5 times.
Sources
- Хабр: OpenAI выпустили GPT-6.2 Astra и Sol и снизили цену флагмана на 20% media outlet
- Kod.ru: Втрое дешевле: представлены GPT-6.2 Astra и Sol media outlet
- 3DNews: OpenAI представила две «жутко умные» модели ИИ — GPT-6.2 Astra для профи и GPT-6.2 Sol для повседневных дел media outlet
- iXBT.com: OpenAI представила GPT-6.2 Astra и Sol: новые флагманские модели стали мощнее и дешевле media outlet
This article was prepared with the help of AI and translated from the Russian original. Spotted an error — let us know.
Published 18 November 2026, 10:01 · No updates · Russian original




Comments on this story live under its post in our Telegram channel. Comment on Telegram