OpenAI Launches GPT 6.2 Astra and GPT 6.2 Sol for Work and Daily Tasks

OpenAI Launches GPT 6.2 Astra and GPT 6.2 Sol for Work and Daily Tasks

OpenAI has launched GPT 6.2 Astra and GPT 6.2 Sol to split heavy professional work from everyday tasks. Both models finish multistep workflows with higher accuracy while burning fewer output tokens. Subscribers and API developers can use both releases right away, with API pricing starting at $2 per 1,000,000 input tokens for Sol and $8 for Astra.

GPT 6.2 Astra handles complex office projects like financial workbooks, slide decks, and legal briefs. OpenAI tuned the model to follow strict corporate templates so workers spend less time fixing drafts. It also writes cleaner code changes and controls CAD software down to individual mechanical parts and joints.

Benchmarks back up these gains. On Agents' Last Exam, Astra beats GPT 6 Astra by 5.2 points at under 33% of the cost, and tops Claude Fable 5.2 by 5.5 points at 5% of the cost. Science testing shows a 9 percentage point jump over GPT 6 Astra on Terminal Bench Science 0.1 alongside a top score on the Vals Science Index. In 3D browser rendering, Astra hits 1318 Elo to take 1st place on AA 3D, passing Claude Fable 5.2 and GPT 6.2 Sol.

GPT 6.2 Sol replaces GPT 6.1 Sol as the daily workhorse inside OpenAI. It is quick. The model writes naturally and pushes through multistep tasks without draining paid usage limits. On Cognition's FrontierCode 2.0 coding test, Sol trails Astra by just 0.5 points while running at roughly 20% of the cost. It also beats the older GPT 6 Astra on the LifeVal daily task benchmark while costing about 7 times less.

Both models cross OpenAI's Critical threshold for cybersecurity and High threshold for biological and chemical risks. They spot software flaws faster and write stronger patches, so OpenAI tightened safeguards to block misuse. GPT 6.2 Astra also reaches the High threshold for AI Self Improvement, meaning it can speed up internal AI research. OpenAI deployed a monitoring system called Control Room to halt unauthorized runs, and updated Codex rules so the models stop hiding missing files or broken tools.

Plus, Pro, Enterprise, and Edu members can run both models in ChatGPT Work and Codex. Pro 500 subscribers get Ultrafast speeds on both models, while Pro 200 users get Ultrafast access on Sol. On the OpenAI API, GPT 6.2 Astra cuts input and output rates by 20% compared to GPT 6 Astra, charging $8 per 1,000,000 input tokens, $0.40 for cached input, and $40 for output. GPT 6.2 Sol keeps existing rates at $2 per 1,000,000 input tokens, $0.10 for cached input, and $10 for output.

About the author

mgtid
mgtid
Owner of Technetbook | 10+ Years of Expertise in Technology | Seasoned Writer, Designer, and Programmer | Specialist in In-Depth Tech Reviews and Industry Insights | Passionate about Driving Innovation and Educating the Tech Community Technetbook

Join the conversation

Newsletter Subscription