Competition in the field of artificial intelligence has once again seen substantive progress. On November 17 local time, OpenAI officially released two fully upgraded flagship models—GPT-6.2 Astra and GPT-6.2 Sol. As a new generation of intelligent engines built for different application scenarios, these two models have achieved significant optimizations in professional work, everyday task handling, and API usage costs.
Dual-Core Drive: Astra Pushes Professional Depth, Sol Leads in Everyday Efficiency
In terms of product positioning and role division, OpenAI uses Astra and Sol to meet users' core needs across different scenarios:
GPT-6.2 Astra (for complex professional tasks): This model has been specifically enhanced for high-difficulty scenarios such as document processing, presentation creation, spreadsheet analysis, software programming, and scientific research. In multiple professional work and scientific research benchmarks, Astra set new score records, demonstrating top-tier logical reasoning and content generation capabilities.
GPT-6.2 Sol (for everyday general-purpose use): As an upgraded version of the previous-generation GPT-6.1 Sol, Sol has become smarter while remaining efficient. It helps users complete more everyday tasks within the same paid plan usage allowance, making it an ideal choice for high-frequency, general-purpose work.
Major Cost Optimization: API Prices Cut Significantly, Ecosystem Fully Open
In terms of the usage costs and ecosystem access that users care about most, OpenAI has also made some highly welcome changes this time:
API prices drop across the board: Thanks to optimizations in its underlying architecture, GPT-6.2 Astra has achieved a clear reduction in the cost of processing tasks. Its API input and output token prices are 20% lower than those of the previous-generation GPT-6 Astra, while cached input prices have been slashed by 60%. Meanwhile, Sol continues to follow the previous GPT-6.1 Sol API pricing standard.
Available on multiple platforms at once: Starting today, both new models are fully available to Plus, Pro, Enterprise, and Edu users of ChatGPT Work and Codex, and developers can also call them through OpenAI's official API.
Security Safeguards Upgraded Again: Reaching "Critical" and "High" Level Defenses
As model capabilities continue to expand, safety and compliance remain a core focus of the industry. OpenAI specifically emphasized in the release that the GPT-6.2 series has further improved its cybersecurity protection capabilities, with all models reaching the "Critical" level for cybersecurity and the "High" level for biological and chemical capabilities in its Preparedness Framework. Based on these rigorous evaluation results, the company has also further strengthened related safety restrictions behind the scenes to ensure the technology remains safe, controllable, and compliant in use.
Join Now