OpenAI releases GPT-6.2 Astra and Sol for professional workflows

OpenAI has introduced GPT-6.2 Astra, positioned for complex professional projects, and GPT-6.2 Sol, an efficient general-purpose model for everyday work. Both are available from today to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers through the OpenAI API.
For Astra, standard API pricing is $8 per million input tokens, $0.40 per million cached input tokens and $40 per million output tokens. OpenAI says these input and output prices are 20% lower than GPT-6 Astra’s, while cached input receives a 60% discount. Sol retains GPT-6.1 Sol pricing: $2 for input, $0.10 for cached input and $10 for output per million tokens.
Two models for complex and routine work
OpenAI describes Astra as its frontier model for demanding work involving slides, spreadsheets, documents, coding, web design, computer use and scientific tasks. It says the model produces outputs that follow supplied templates and professional conventions more closely, with more direct writing and fewer generic takeaways.
On Agents’ Last Exam, OpenAI reports that Astra scored 5.2 points above GPT-6 Astra at less than one-third of the cost. The evaluation tests long-horizon, multistep professional workflows. The company also says Astra scored nine percentage points higher than GPT-6 Astra on Terminal-Bench-Science 0.1 at less than one-third of the cost.
Sol is intended as the default option for technical and non-technical work. On Cognition’s FrontierCode 2.0, OpenAI says Sol came within 0.5 points of Astra at about one-fifth of the cost when both used maximum settings. The benchmark assesses whether changes to real codebases satisfy maintainers’ requirements for merging.
Document, design and engineering claims
OpenAI trained Astra for work in law, sales, finance and accounting, including the production of briefs, presentations and workbooks. On its internal GDPval v2 benchmark, the company says Astra improved by 6.2 percentage points over GPT-6 Astra while using 22% fewer output tokens; Sol improved 5.3 points over GPT-6.1 Sol.
The release also emphasizes visual reasoning and computer use. OpenAI says Astra can work with CAD and simulation tools at the level of individual parts and connections, and can build interactive browser experiences from written briefs. In Artificial Analysis’s AA-3D evaluation, Astra ranked first at 1,318 Elo, ahead of Claude Fable 5.2 at 1,271 and Sol at 1,259.
The pricing progression follows GPT-6.1 Sol’s fivefold price gap with Astra and now places Astra’s lower token rates alongside a new model generation aimed at complex professional workflows.
Cybersecurity safeguards and deployment
OpenAI says both GPT-6.2 models meet the Critical threshold for cybersecurity under its Preparedness Framework. The company is deploying tighter safeguards for higher-risk cybersecurity assistance, noting that the restrictions can also limit some legitimate vulnerability research and security-testing workflows.
Both models also meet the High threshold for biological and chemical capabilities, OpenAI says. Astra is the first externally released OpenAI model to meet the High threshold for AI self-improvement. The company says it has added oversight, containment and monitoring controls, including a system called Control Room that can halt further model output when intervention is needed.
For businesses, the practical implication is to test Astra and Sol against real document, coding and agent workflows, select the model that meets the required quality level, and keep prompt prefixes unchanged when possible to improve cache reuse, latency and token costs.

