GPT-5.6 Sol matches Mythos Preview on ExploitBench, adds Ultra mode with subagents for complex workflows, and max reasoning for deep problem-solving (OpenAI)

OpenAI has launched GPT-5.6 Sol with advanced capabilities including Ultra mode with subagents for orchestrating complex workflows and enhanced reasoning for deep problem-solving, positioning it competitively with existing benchmarks while offering a three-tier model lineup (Sol, Terra, Luna) to address varying organizational needs and cost-performance tradeoffs. For IT leaders, this expansion enables more sophisticated automation of enterprise workflows, improved decision support for complex technical challenges, and opportunities to optimize AI spending across different use cases. The subagent architecture particularly impacts IT operations by enabling autonomous coordination of multiple AI agents for infrastructure management, security analysis, and software development tasks.