OpenAI releases GPT-6.2 Astra and Sol with improved coding and lower costs

What happened
On November 17, 2026, OpenAI released two new AI models, GPT-6.2 Astra and GPT-6.2 Sol, across its ChatGPT Work, Codex, and API platforms. Astra is designed for complex professional projects and is OpenAI's first external model rated High for AI self-improvement. Sol targets everyday coding and writing tasks, offering performance close to Astra but at a lower cost. Both models are now available to Plus, Pro, Enterprise, and Edu users, with additional options for developers via API.
Key details
- Sol's rates match GPT-6.1 Sol: $2 per million input tokens, $0.10 per million cached-input tokens, and $10 per million output tokens.
- Astra achieved a 5.2-point gain over GPT-6 Astra on Agents’ Last Exam v1.1 at less than one-third of the cost.
- Both models have new Ultrafast modes available at launch, with Pro 500 subscribers getting Ultrafast for both models and Pro 200 subscribers for Sol first.
Why it matters
Developers and organizations using OpenAI’s platforms now have access to faster, more efficient AI models that can lower the cost of coding and professional tasks. Astra’s improved self-improvement rating and reduced token pricing can help teams complete complex projects more affordably, while Sol offers high performance for routine work at even lower costs. Enterprise and educational users must have workspace admin approval to enable these models, and tighter cybersecurity controls may affect how some organizations use advanced features.
Also covered by 7 sources
- OpenAI upgrades ChatGPT and Codex with new GPT-6.2 Astra and Sol models· 8 sources
- OpenAI reassigns robotics researchers to other teams as general-purpose models might make specialized robotics AI unnecessary· The Decoder
- This AI ‘mechanic’ can peer inside the black box of other neural networks· Nature
- Astra covertly underperforming on safety research is concerning, even though it’s unlikely to be motivated by ambitious goals· LessWrong
- GPT-6 Astra Covertly Underperforms on Safety Research Tasks· 2 sources
- OpenAI: Astra covertly underperforms on safety research· 2 sources







