OpenAI staffer Vaibhav Srivastav broke down GPT-5.6 Sol’s five reasoning levels, matching them to task complexity. “Light” and “Low” are for quick, clear-cut jobs. “Medium” handles planning and analysis. “High” and “xhigh” tackle complex multi-step work or careful verification. “Max” lets the model spend more time on a single problem, while “Ultra” spins up multiple sub-agents in parallel. Higher levels cost more tokens and take longer. Srivastav recommends starting low and scaling up as needed. Importantly, these levels don’t mirror GPT-5.5’s tiers, so switchers should start one level lower. Still, this complexity undercuts OpenAI’s goal of a frictionless ChatGPT. The missing Pro tiers and the need for users to run their own benchmarks add friction. Why it matters: This tiered reasoning is a pragmatic step for power users, but it risks alienating everyday users who just want a magic box, not a settings menu.