OpenAI fixed GPT-5.6 Sol’s most frustrating flaw: Burning limits while it waits
OpenAI addressed a significant user complaint in GPT-5.6 Sol: usage limits were being consumed while the model waited for input. The issue, described as "burning limits while it waits," meant that users lost quota during idle periods, such as when the model paused mid-conversation. The fix ensures that usage limits reset when the model is idle, so limits only decrement during active processing. This change directly responds to user frustration over wasted capacity, making the system more efficient for developers who rely on extended interactions. The update applies to GPT-5.6 Sol, the latest model iteration, and is now live. No other changes to the model's capabilities or pricing were announced.
Developers no longer lose usage quota during idle model pauses, improving cost efficiency.