OpenAI GPT-5.6: Advanced Reasoning, Cybersecurity Unveiled

OpenAI has launched its new GPT-5.6 models, featuring advanced reasoning and enhanced cybersecurity. These models come in three tiers: Sol, Terra, and Luna. GPT-5.6 Sol is the most capable, with a "max" reasoning effort for deeper analysis and an "ultra" mode for complex workflows. It shows strong performance in command-line planning and genomics tasks. For cybersecurity, GPT-5.6 Sol improves vulnerability research and uses fewer output tokens compared to other models. OpenAI built in robust, layered safety protections. They used over 700,000 GPU hours for automated red-teaming to find and fix potential vulnerabilities before launch. The models are currently in a limited preview with broader availability coming soon through ChatGPT, Codex, and the API. This development could bring more powerful and secure AI tools to a wider audience.