Key facts
- OpenAI said GPT-6.2 Astra scored 5.2 percentage points above GPT-6 Astra on Agents' Last Exam.
- OpenAI said Astra beat Claude Fable 5.2 by 5.5 points on that test and cost 5% less per task.
- OpenAI said Astra scored nine percentage points above GPT-6 Astra on Terminal-Bench-Science 0.1.
- OpenAI said Sol runs near Astra on Cognition's FrontierCode 2.0 at roughly one-fifth of the price.
- OpenAI cut GPT-6.2 Astra API token prices by 20% from GPT-6 Astra levels.
- OpenAI said both models meet its Critical cybersecurity threshold and High biological and chemical threshold.
OpenAI launched two models, GPT-6.2 Astra and GPT-6.2 Sol, on November 17, 2026, saying both improve reasoning, coding, scientific research and professional workflows while reducing processing costs. Astra is aimed at complex assignments and Sol at everyday work, the company said.
OpenAI said Astra performed 5.2 percentage points better than GPT-6 Astra on Agents' Last Exam, a test of multistep professional tasks, and beat Claude Fable 5.2 by 5.5 points while costing 5% less per task. It said Astra scored nine percentage points above GPT-6 Astra on Terminal-Bench-Science 0.1 and set a new high on the Vals Science Index, reflecting performance in analyzing biological data and workflow research.
Sol, which OpenAI said is optimized for common professional and technical problems, performs near Astra on Cognition's FrontierCode 2.0 at roughly one-fifth of the price, the company said. OpenAI said Sol also beat GPT-6 Astra on its LifeVal evaluation of real-world daily tasks at roughly one-seventh of the price.
OpenAI cut GPT-6.2 Astra API token prices by 20% from GPT-6 Astra levels, to $8 per million input tokens, $0.40 per million cached tokens and $40 per million output tokens. Sol keeps the $2 per million input and $10 per million output rates of its predecessor, the company said. The pricing update follows OpenAI's October partnership with Google Cloud to sell its models to enterprise customers.
Both models are available to Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex, and to developers through the OpenAI API with Ultrafast processing at launch. OpenAI said it tightened safeguards because the models show stronger cybersecurity capabilities, restricting certain high-risk cybersecurity requests while planning to support approved defensive research through its Daybreak program. The launch comes as SpaceXAI has released Grok 4.7 and Anthropic continues developing competing frontier models.