GPT-5.6 Users Report Rapidly Depleting Usage Limits

Following the release of GPT-5.6, numerous users reported that the model depletes subscription usage limits much faster than before, sparking widespread frustration in the community. Several ChatGPT Plus and Pro users noted that the new model exhausts quotas and triggers limits extremely quickly when handling complex tasks or using high-compute tiers, contrasting sharply with official claims of being more token-efficient.

User Reactions and Specific Feedback

Users encountered quota bottlenecks in various scenarios. User @skrr2 reported that as a Plus subscriber, their usage limit was rapidly exhausted after performing just two tasks: merging about 10 PDFs into a 700-page document and organizing around 700 Obsidian notes. User @koltregaskes pointed out that using the Extra High tier of GPT-5.6 Sol triggered a 5-hour limit after just a few messages, despite having a Pro x5 subscription that normally allows for more usage. Additionally, users noted that version 5.6 consumes limits faster than 5.5 in scenarios like Codex.

Controversies and Speculations

Regarding the cause of the rapid quota depletion, the community offered different perspectives. Cited reports mentioned that Sam Altman claimed GPT-5.6 Sol achieves a 54% higher improvement in agentic coding tasks, and the new version appears to be more token-efficient. User @ssh4net questioned how being "more token-efficient" and "depleting limits faster" could be true simultaneously. User @dejavucoder speculated that the faster exhaustion in sol xhigh fast mode might be related to the new version's tendency to invoke sub-agents, though this remains a guess. To cope with the usage limits, the community currently suggests lowering the reasoning effort.

2026-07-10 ~ 2026-07-12 · 6 related posts

Full story(20 episodes)→