AI3 min
OpenAI's GPT-6.2 Astra ranks near the top in outside tests but is no better at coding
OpenAI introduced GPT-6.2 Astra, its frontier model for hard projects, and GPT-6.2 Sol, an efficient model for everyday work. Both are in ChatGPT Work, Codex and the API, for Plus, Pro, Enterprise and Edu users. OpenAI says Astra sets records on work and science tests at under a third of GPT-6 Astra's cost. According to 9to5Mac, Astra costs $8 per million input tokens and $40 per million output tokens, 20% less than GPT-6 Astra, while Sol keeps its $2 and $10 prices. The first independent rankings are in. According to The Decoder, Artificial Analysis puts Astra second in its main index, behind only Anthropic's Claude Fable 5.2, and level with GPT-6 Astra in coding. Epoch AI has Astra tied with Fable 5.2. GitHub says both models are now in Copilot. Reviewer @theo called Astra frustrating: he says it has no sense of what it does not know, its outputs are hard to read, and it runs in circles on long tasks.
Hype check: Outside rankings back OpenAI's claim that Astra is near the top, but not that it is a clear leap, especially for coding. If you already pay for ChatGPT or Copilot, try it on your own work before switching anything.