← sources

How to reduce token usage for PDF output?

source: reddit (reddit:ClaudeCode)published 2026-10-24 13:41ingested 2026-10-25 02:39

retained text

I will start by saying I am Claude newbie and I am absolutely not a developer by any stretch. I am using Claude to help with my job search. I found a GitHub repo designed for job search. I actually had to have Claude walk me through step by step how to set that up. Including creating a GitHub account. So needless to say I barely know what I’m doing. 

With the repo installed I can scrape a bunch of job postings. It will then rank the finding based on a profile we have established. Then it will create a PDF resume based on that profile and past resumes. It works pretty well actually. 

I will usually do the scrape and rank process using Sonnet to save tokens then move over to a session with Opus to actually build the resume. I realize Opus uses considerably more tokens than Sonnet. 

I always see repos talking about reducing token usage. I’m not really “building” much here other than the PDF resume so I didn’t know if they would actually apply here. Are there any that anyone can recommend that would be a good fit to reduce tokens when mostly during that Opus session that’s creating the resume? 

document fetched 2026-10-25 00:35 · canonical https://reddit.com/r/ClaudeCode/comments/1xe8z7x/how_to_reduce_token_usage_for_pdf_output