Image: iGeekPhone / openai.com OpenAI Temporarily Lifts the 5-Hour Cap on Codex and ChatGPT Work
OpenAI removed the rolling 5-hour usage limit for Plus, Business, and Pro subscribers on July 12, after demand from GPT-5.6 Sol drained usage budgets faster than expected.
OpenAI lifted the five-hour rolling usage cap on Codex and ChatGPT Work on July 12. The change applies to Plus, Business, and Pro subscribers, and it’s explicitly temporary — no end date attached.
The five-hour window was a rolling limit that cut off access once a user hit it, separate from any weekly cap. With it gone, the only remaining ceiling is the weekly total, which stays in place.
Why now
The timing follows the broad rollout of GPT-5.6 Sol on July 11. Sol is the top reasoning tier in the new GPT-5.6 family, and it burns through compute faster than what users were used to. OpenAI’s engineering lead for Codex, Thibault Sottiaux, noted that usage budgets were resetting within an hour of the announcement, acknowledging that the demand spike caught the team off guard.
The company also disclosed that the platform has crossed 6 million active users — context for how quickly load compounds when a new high-compute model goes broad.
Sol efficiency improvements coming
OpenAI said it’s working on inference efficiency improvements for Sol specifically. The stated goal is to let the same weekly budget get users further without reducing what they can do with it. No timeline on when that lands.
Until then, the lift buys breathing room. Users who hit the five-hour wall repeatedly can now work through longer sessions without forced interruption.
What’s still limited
The weekly usage cap is still there. OpenAI hasn’t changed the per-plan total allocations, so heavy users will still run into limits before the week resets — they just won’t hit the shorter intra-week wall first.
The removal also covers ChatGPT Work, the agent mode launched alongside the unified desktop app on July 9. Work runs on GPT-5.6, with Pro and Enterprise accounts getting Sol and lower-tier plans getting Terra.
What to expect next
OpenAI has been clear this is a temporary window, not a policy change. When limits return, they’ll likely reflect whatever efficiency improvements land in the meantime. The safest read is that OpenAI is buying time to tune Sol’s compute footprint before reintroducing restrictions.
Sources: iGeekPhone · Digital Trends · BleepingComputer