Weekly limit dropping as fast as 5-hr limit

GPT-5.5 saves time by doing more work with fewer tokens. However, alongside that, it can consume up to 2x more credits compared to GPT-5.4. If you are a Plus subscriber, this is a fairly normal usage rate. With a Pro subscription, I remember hitting around 25–30% of my rate limit in one day using “GPT-5.5 extra-high.” By using .md documentation files and keeping those documents up to date, you can help the model spend less of your rate limit. This way, whenever you want to make a change, it can reach the target with more precision and less effort, giving your rate limit a longer lifespan. I do not know your working style or documentation discipline, since you did not provide details.