OpenAI Introduces GPT-6.2 Astra And Sol With Improved Performance And Lower Task Costs

OpenAI has introduced GPT-6.2 Astra and GPT-6.2 Sol, expanding its model lineup with improvements in professional work, coding, scientific research and token efficiency.

Astra is positioned for complex projects, while Sol targets everyday technical and nontechnical tasks. Both models are available through ChatGPT Work, Codex and the OpenAI API.

OpenAI said Astra improves the quality of documents, spreadsheets and presentations, including closer adherence to supplied templates and professional conventions. The company also reported gains in software development, interactive web design, computer use and three-dimensional applications.

On Agents’ Last Exam, which evaluates multistep professional workflows, Astra scored 5.2 percentage points above GPT-6 Astra at less than one-third of the cost, according to OpenAI.

On its internal GDPval v2 evaluation, Astra improved by 6.2 percentage points while using 22% fewer output tokens. Sol improved by 5.3 points over GPT-6.1 Sol.

Sol also came within 0.5 points of Astra on Cognition’s FrontierCode 2.0 coding evaluation at approximately one-fifth the cost when both used maximum reasoning.

API Pricing And Availability

Standard Astra pricing is $8 per million input tokens, $0.40 per million cached input tokens and $40 per million output tokens. Input and output rates are 20% below GPT-6 Astra’s, while cached input pricing is 60% lower.

Sol retains its predecessor’s rates of $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens.

The models are available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. Eligible Enterprise and Edu access requires workspace administrator activation.

OpenAI is also extending Ultrafast access to Pro 200 subscribers, initially for Sol. Pro 500 subscribers can use Ultrafast with either model, and developers have access to it for both models through the API.

Scientific Capabilities And Safeguards

OpenAI reported improved performance on scientific research evaluations, including a nine-percentage-point gain over GPT-6 Astra on Terminal-Bench-Science 0.1 at less than one-third the cost.

Astra is also the company’s first externally released model to reach its High threshold for AI self-improvement, a classification assessing the ability to accelerate AI research.

Both models meet OpenAI’s Critical cybersecurity threshold and High biological and chemical capability threshold. The company said it strengthened restrictions, monitoring and intervention systems for higher-risk applications.

OpenAI also disclosed that the models were more likely to omit work-blocking obstacles in two evaluations. It said additional Codex instructions substantially reduced those omissions, with further investigation continuing.

KEY QUOTES:

“We evaluated GPT-6.2 Astra on turning dense information into readable briefs and infographics for both technical and non-technical audiences. Examples included trading briefs and diagrams of agentic workflows.

GPT-6.2 Astra’s documents were a significant step up from GPT-6 Astra. The writing and charts made it much easier to pick out the key points of a briefing, and the diagrams clarified complex systems by showing how the parts work together.”

Marquis Wong, Principal AI Engineer At IMC

“GPT-6.2 Astra is a huge leap forward for creative work. It makes Higgsfield Miracle possible, powering real-time video editing with pixel-level precision and giving creators ultimate creative control. In our internal 3D generation evals, it achieved the target result in one shot in 73% of cases, while GPT-6 Astra needed three or more iterations. This saves our creative team a lot of time and allows for more efficient fully agentic workflows.”

Alex Mashrabov, CEO And Co-Founder Of Higgsfield AI

“GPT-6.2 Astra has a huge improvement over GPT-6 Astra in token and tool efficiency by better planning long trajectories and more efficiently writing Bash to do more in fewer turns. Our average time for long-horizon accounting work fell by 50%, with the time for the most complex work falling up to 73%.”

Eros Wang, Member Of Technical Staff At Basis