OpenAI ships GPT-6.2 Astra and Sol with mixed early reviews
Original title: [AINews] GPT-6.2 Astra, the #2 Frontier Model is better and 20%-600% cheaper, but has mixed reviews; 6.2 Sol likely main workhorse
Astra is $8/$40 per million tokens; Sol stays at $2/$10, and both offer 1M context.
OpenAI released GPT-6.2 Astra and Sol to ChatGPT Work, Codex and the API as gpt-6.2-astra and gpt-6.2-sol. Artificial Analysis scored new Astra at 62 on its Intelligence Index, behind Fable 5.2 at 65, while Sol ties September’s Astra at about 15% of the cost per task. OpenAI rated Astra High for AI Self-Improvement. Early testers disagreed, with Matt Shumer calling the models essentially unusable after weeks of testing.
Unconfirmed: Early tester criticism is disputed by the newsletter, which still treats some harness and RL complaints as valid.
- Astra API prices are $8 input, $0.40 cached input and $40 output per million tokens.
- Sol keeps prior pricing at $2, $0.10 and $10; both have a 128K output limit.
- Epoch verified seven Lean solutions out of 68 FrontierMath Erdős problems, up from three.
- Enterprise Astra adds a legal search index and scored 20.8% on Harvey LAB-AA versus 15.0% for base Astra.
- Vendors
- OpenAI
- Organizations
- Artificial AnalysisAnthropicEpochVals
- AI models
- GPT-6.2 AstraGPT-6.2 SolFable 5.2
From the original article
6 min read at latent.space
November 17, 2026 — GPT-6. 2 Astra and Sol lower benchmark costs. Early testers disagree on what Astra is like to work with.
By swyx .ai · latent.space
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.