Image: 9to5Mac / 9to5mac.com OpenAI Boosts GPT-5.6 Sol Efficiency 18%, Restores Codex 5-Hour Limit July 30
After temporarily lifting Codex's rolling usage cap during the GPT-5.6 Sol surge, OpenAI is bringing it back Thursday with a Sol efficiency improvement meant to offset the tighter window.
Tibo Sottiaux, the Codex lead at OpenAI, posted another limit update on July 29: a fresh usage reset for Work and Codex subscribers, an efficiency improvement to GPT-5.6 Sol that should stretch each session about 18% further, and a heads-up that the five-hour rolling cap returns on July 30.
The five-hour limit had been suspended since July 12, when demand for GPT-5.6 Sol doubled traffic within 48 hours of launch. At the time, OpenAI lifted the rolling window for Plus, Pro, and Business subscribers, reset usage counters, and pushed an initial Sol efficiency improvement. The five-hour cap was always described as temporary, but the pairing with efficiency improvements suggests OpenAI wants the reintroduction to feel less abrupt.
What’s Changing
The five-hour limit worked as a rolling window — not a daily reset. If you started a long session, the clock started from your first message. With the cap suspended, the only ceiling was the weekly limit shared between Chat and Codex tasks.
The reinstated window comes back July 30 with Sol running more efficiently underneath. OpenAI hasn’t published the exact mechanism for the 18% figure, but the effect should be fewer credits burned per task at the same output quality. In practice, that translates to more work fitting inside a five-hour window than the same window allowed before July 12.
The shared credit pool between ChatGPT Work and Codex stays the same. Requests from cloud tasks and local messages all count against the same bucket.
Context on the Sol Launch
GPT-5.6 Sol launched on July 9 across ChatGPT, Codex, and the API. It’s OpenAI’s flagship for coding and reasoning work, priced at $5 input / $30 output per million tokens. By July 14, Codex and ChatGPT Work had reached 8 million active users combined — up from around 1 million in February — which explains why the infrastructure team was scrambling to manage limits in the first place.
The temporary suspension was a reasonable emergency valve. Bringing back the limit with a measurable efficiency improvement is more orderly than a cold restoration, though the exact impact on real workflows depends on how Sol-heavy your sessions actually run.
If you’re on Plus, Pro, or Business and were using the limit-free window for longer uninterrupted agent runs, tomorrow’s the last day before the five-hour cap reapplies.
Sources: