← sources

Claude Pro usage: 0% to 100% in one prompt

source: reddit (reddit:ClaudeAI)published 2026-11-11 19:23ingested 2026-11-12 02:37

retained text

First use of the day on the Pro plan. I sent one prompt using Fable 5.2 at medium effort.

I asked it to compare three project timeline documents and identify strengths and potential failure points. The request didn’t complete, and my session usage was maxed out. I got no completed review.

EDIT: The prompt was

>*"I created a launch timeline and an email launch plan based on DOCUMENT\_A.docx. Please evaluate their strengths and weaknesses. DOCUMENT\_A.docx is the original campaign timeline. DOCUMENT\_B.xlsx is my launch timeline with weighted line items and estimated task hours. DOCUMENT\_C.docx is my email launch plan. I created the latter two to make the plan more effective and provide a clearer picture of the workload."*

The files:

* DOCUMENT\_A.docx: the original campaign timeline, approximately 86 KiB. It’s protected with Office rights management, so I don’t know whether Claude could read it or how it handled the protection.
* DOCUMENT\_B.xlsx: the weighted launch timeline, approximately 119 KiB. It has five tabs and about 9,900 populated cells, including roughly 4,100 mostly repeated planning formulas.
* DOCUMENT\_C.docx: the email launch plan, approximately 40 KiB, with about 880 words and three tables.

The files total roughly 245 KiB. Compressed file size doesn’t tell us how much content or processing the model uses, and I can’t tell from the usage meter what caused the limit to be exhausted. Neither attempt produced a completed review or explained an issue with the protected file.

During the retry, I saw the usage meter at 2% while Claude was reasoning. It jumped to 100% within seconds, too quickly for me to capture the change. The before and after screenshots include timestamps. I’ve redacted most of the surrounding UI for privacy because the work involves a five-product launch under embargo.

I used a full reset and tried again with the exact same prompt, documents, and settings. Same result: the request didn’t finish, and the session went from 0% to 100% used. That left the account unusable for the remainder of my workday.

The attached screenshots show the session at 100% before the reset, 0% after the reset, and 100% after the retry. The first and last show 12:02 p.m. and 12:13 p.m. The reset screenshot doesn’t have a visible clock. I redacted account details, unrelated conversations, and work information for privacy; the usage figures remain visible.

**This happened after I downgraded from the $100/month plan to the $20/month plan because of poor model performance and an argumentative tone.**

By "argumentative tone", I mean repeated lectures despite preferences stated in my account settings and reinforced through memory requests, plus abrupt changes in tone when I challenge its reasoning or findings.

You can blame the prompt all day long, but when I can use a competing or open-source product and get higher-quality results with simpler language, the failure isn’t on the consumer. It’s on the provider.

Has anyone else had a single request burn through an entire session without completing? Is this expected with document comparisons, or does it look like something went wrong?

document fetched 2026-11-12 00:22 · canonical https://reddit.com/r/ClaudeAI/comments/1xtr0e3/claude_pro_usage_0_to_100_in_one_prompt