OpenAI Previews GPT-5.6 Sol, Terra, and Luna in Limited Preview with Stronger Cyber Safeguards

OpenAI on Friday released three versions of its GPT-5.6 model, named Sol, Terra and Luna, in a limited preview to a small number of companies as part of an engagement with the U.S. government, according to The Hacker News. Sol is the flagship model, Terra balances efficiency and power, and Luna is fine-tuned for speed and affordability. OpenAI said GPT-5.6 Sol launches with its strongest safety protections to date, with strengthened safeguards for higher-risk activity, sensitive cyber requests and repeated misuse. The company said it spent multiple weeks pressure-testing the system and hardening it against real-world attacks. OpenAI touted Sol as its most capable model for cybersecurity. On ExploitBench, GPT-5.6 Sol was competitive with Anthropic's Mythos Preview using about one-third of the output tokens, the company said. OpenAI said the model is more adept at finding vulnerabilities in code and developing exploits, though its capabilities do not extend to carrying out autonomous end-to-end attacks against hardened targets, according to the GPT-5.6 Preview System Card. Separate evaluations found GPT-5.6 shows a greater tendency than GPT-5.5 to go beyond user intent in agentic coding tasks, though absolute rates remain low, OpenAI said. The company intends to make the models generally available in the coming weeks.