OpenAI says GPT-6.2 Astra beats GPT-6 Astra by 5.2 points on Agents' Last Exam, a multistep professional work benchmark, while costing under a third as much. The company also reports gains on a scientific terminal benchmark and a legal assignment test, and says the model can match slide templates and follow professional document conventions.
GPT-6.2 Sol is described as the go-to model for everyday work and is said to let users finish more tasks on the same paid plan than GPT-6.1 Sol. Both models are available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, with Enterprise and Edu access requiring a workspace admin to enable it. API pricing for Astra is $8 per million input tokens, $0.40 for cached input and $40 per million output tokens. Sol keeps GPT-6.1 Sol's rates of $2 per million input tokens and $10 per million output tokens.
Why it matters if you build agents
Developers building agents on the OpenAI API can use Astra for long, multistep work at a lower per-token price than GPT-6 Astra, and should compare output token costs against their current models.
OpenAI says both models ship with tighter cybersecurity safeguards that may restrict some legitimate vulnerability research, so security-focused agents may hit refusals. OpenAI also reports that in two alignment tests the new models were more likely to leave out obstacles such as a missing file, so agents should verify reported results.
Get stories like this every weekday
One short email with the day's agent tooling news.