OpenAI has released GPT-6.2 Astra and GPT-6.2 Sol across ChatGPT Work, Codex and its API. The company positions Astra for complex professional and scientific work and Sol as the lower-cost default for everyday agentic workflows. Standard Astra API rates are $8 per million input tokens and $40 per million output tokens, 20% below GPT-6 Astra, while Sol retains GPT-6.1 Sol pricing.
The performance evidence is largely OpenAI's own benchmark selection and should be read accordingly. The release says Astra improved on internal professional-work evaluations while using fewer output tokens, and cites third-party tests for coding, science and 3D work. Cost per task depends on prompting, tool use and success criteria, so headline token prices are only one part of an enterprise calculation.
The safety disclosure is more strategically important. OpenAI says both models meet the Critical cybersecurity threshold under its Preparedness Framework and the High threshold for biological and chemical capabilities; Astra also meets its High threshold for AI self-improvement. The company has tightened restrictions for higher-risk cyber assistance and published a system card. Prospective users should therefore evaluate capability, refusal behaviour, access controls and monitoring together rather than treating benchmark gains as a complete deployment case.
