← sources

High token consumption on Claude just for context analysis: any tips to optimize?

source: reddit (reddit:ClaudeAI)published 2026-11-19 08:25ingested 2026-11-20 02:36

retained text

Hi everyone,

I’m looking for a bit of beginner advice on using Claude.

I’m currently working on a project, and I’ve noticed that the system uses up a massive amount of tokens on every single interaction just to analyze the context. Even when I send a minimal prompt that doesn't require any heavy lifting or complex analysis, the token usage remains really high.

I'm currently using the base model for this workflow. Based on your experience, are there specific skills, guides, or best practices I can use to keep this consumption under control and make my workflow more efficient?

Thanks a lot in advance to anyone who can help out!

document fetched 2026-11-20 00:23 · canonical https://reddit.com/r/ClaudeAI/comments/1y04ac7/high_token_consumption_on_claude_just_for_context