OpenAI Previews GPT-5.6 Sol With Restricted Access and Stronger Cyber Safeguards

OpenAI released GPT-5.6 Sol, the most powerful of three new versions, as a limited preview to select companies under U.S. government engagement. The model matches Anthropic Mythos Preview on ExploitBench using a third of the output tokens and is touted as the most capable for cybersecurity. Safeguards block offensive activity and include enhanced protections against jailbreaking.