TechFlow reported that on November 18, the B.AI platform announced the official launch of OpenAI's latest GPT-6.2 series reasoning models, GPT-6.2 Astra and GPT-6.2 Sol, with simultaneous availability on both API and Web Chat. The entire lineup features a 1.05M token ultra-long context window, adjustable reasoning intensity ranging from low to max, multimodal image-text understanding, and supports computer operation and multi-agent task delegation. GPT-6.2 Astra is designed for high-complexity software engineering and in-depth professional projects, featuring capabilities in cross-component code debugging, visual computer operation, and complex system planning; GPT-6.2 Sol targets routine software development, professional office tasks, and agent workflows, offering better call costs while preserving multi-step iterative execution capabilities. Log in to chat.b.ai/chat to try it out.
Add to Favorites
Share to Social Media





