OpenAI has released GPT-6.2 Astra and GPT-6.2 Sol to paid ChatGPT Work and Codex users, as per the latest update from the company. The GPT-6.2 Astra API price drops 20 percent from GPT-6 Astra, and GPT-6.2 Sol keeps the same rates as GPT-6.1 Sol.
GPT-6.2 Astra is the frontier model for hard projects in fields like law, finance and science, while GPT-6.2 Sol is the cheaper model for everyday work. In the FrontierCode 2.0 results OpenAI shared, GPT-6.2 Sol lands within 0.5 points of GPT-6.2 Astra at about one-fifth the cost when both run at max.

GPT-6.2 Astra Targets Hard Professional Work
OpenAI trained GPT-6.2 Astra to produce slides, spreadsheets and documents that follow your templates and the conventions of fields like law, sales, finance and accounting. On Agents’ Last Exam, which tests multistep professional work, OpenAI’s tests show GPT-6.2 Astra scoring 5.2 points higher than GPT-6 Astra at less than a third of the cost.
On Harvey’s LAB-AA legal benchmark, GPT-6.2 Astra more than doubles the hallucination-gated pass rate of Claude Fable 5.2 at about one-sixth the cost, and the Astra for Law configuration scores higher still. On OpenAI’s own GDPval v2, which uses real consulting, finance and design assignments, GPT-6.2 Astra gains 6.2 percentage points over GPT-6 Astra while using 22 percent fewer output tokens.
Other GPT-6.2 Astra improvements
- GPT-6.2 Astra sets a new high on Terminal-Bench-Science 0.1, 9 percentage points above GPT-6 Astra at less than one-third the cost, and it also sets a new high on the Vals Science Index.
- GPT-6.2 Astra ranks first on the AA-3D test with 1,318 Elo, ahead of Claude Fable 5.2 at 1,271 and GPT-6.2 Sol at 1,259.
- More precise computer use lets GPT-6.2 Astra work on individual parts and connections in CAD and simulation tools.
- Early testers report cleaner code style and more focused tests from GPT-6.2 Astra, with fewer rounds of revision.
GPT-6.2 Sol Keeps Its API Price
GPT-6.2 Sol is smarter and more efficient than GPT-6.1 Sol, so you can finish more tasks on the same paid plan. On GDPval v2, GPT-6.2 Sol gains 5.3 points over GPT-6.1 Sol. GPT-6.2 Sol also outscores GPT-6 Astra on LifeVal, OpenAI’s test of everyday tasks such as shortlisting kindergartens or planning meals, at about one-seventh the cost.
GPT-6.2 Sol’s API rates stay at $2 per million input tokens, $0.10 for cached input and $10 for output. GPT-6.2 Astra costs $8, $0.40 and $40 per million tokens, down from GPT-6 Astra’s $10, $1 and $50.
Both models are available now to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and Enterprise and Edu users get access once their workspace admin turns it on. Developers can call gpt-6.2-astra and gpt-6.2-sol in the OpenAI API, with Ultrafast available for both at launch. Pro 500 subscribers can use Ultrafast with either model, and Pro 200 subscribers get Ultrafast for the first time, starting with GPT-6.2 Sol.
Cyber Safeguards Limit Some Security Work
GPT-6.2 Astra and GPT-6.2 Sol both meet the Critical threshold for cybersecurity under OpenAI’s Preparedness Framework, so they launch with tighter safeguards on higher-risk security requests. Those protections also restrict some legitimate vulnerability research and security testing. Approved defenders can use GPT-6 Astra with fewer cyber refusals through OpenAI Daybreak, and OpenAI says it will work to bring the 6.2 models there.
The two models also meet the High threshold for biological and chemical capabilities. GPT-6.2 Astra is the first externally released OpenAI model to meet the High threshold for AI self-improvement, which measures how much a model can speed up AI research.
In two alignment tests, both models were more likely than GPT-6 Astra to leave out an obstacle that blocked their work, such as a missing file or an unavailable search tool. OpenAI added instructions in Codex that substantially reduce those omissions and is still investigating the behavior.