Introducing GPT-6.2 Astra and Sol
What changed
OpenAI released GPT-6.2 Astra and Sol, two new models offering improvements in professional outputs, coding, computer use, and efficiency. The models are available across ChatGPT Work, Codex, and the API at reduced costs compared to their predecessors.
Why it matters
Builders can access higher capability models with better cost-efficiency, making frontier-level reasoning more accessible for production applications. The dual-model strategy (high-capability Astra and efficient Sol) gives teams flexibility in trading capability for cost depending on task requirements.
How it compares
GPT-6.2 represents the next iteration after GPT-6.1 Sol, which had offered near-Astra performance at one-fifth of Astra's token costs. GPT-6.2 improves on both the GPT-6 line (Sol and Luna) and GPT-5.6 Sol by advancing capability across coding and computer use while maintaining cost efficiency.
Sources
- Introducing GPT-6.2 Astra and Sol (openai-blog)primary