OpenAI temporarily lifts GPT-5.6 Sol usage limits amid demand surge

Days after launching GPT-5.6, OpenAI temporarily lifted the five-hour usage restriction on its flagship Sol model for Plus, Pro, and Business subscribers, responding to a surge in demand, and granted a one-time usage reset for all users. In parallel, it began rolling out efficiency improvements to reduce Sol's token consumption so users accomplish more within their allotted limits.

The move is a classic post-launch pressure-release valve: a new flagship draws heavy usage, limits bite, and users complain, so the vendor loosens constraints while quietly optimizing serving costs. The efficiency push — fewer tokens per task — is the durable fix; the limit removal is the temporary goodwill gesture.

Ethan Mollick captured the frustration this creates for planning: he argued that even a rough forward commitment ('we intend to keep extending this week-by-week but may stop under the following conditions') would help enterprises plan AI adoption, criticizing the ad-hoc, temporary nature of the limit changes (1,008 likes). The uncertainty around when limits snap back makes it hard for teams to build workflows on Sol.

The episode also reflects the compute-economics squeeze underlying the whole week: OpenAI can't afford unlimited Sol access at current serving costs, so it's racing to make the model cheaper to run even as it courts users with temporary generosity. Watch whether the efficiency gains let OpenAI make higher limits permanent, or whether the five-hour cap quietly returns once the launch surge subsides.