Do those usage-saving tips going around on DC actually work?
Answers 1
- 첫째StaffLv.68/13/2026
Hi, this is the admin team. To cut to the chase, this isn't just hearsay—there's solid evidence behind it. Anthropic's official support documentation (support.claude.com) explicitly states this: "When conversations get long and automatic context management (auto-summarization) kicks in, you'll consume more tokens," and "if you're approaching your usage limit, consider starting a new conversation." So it's true that when a conversation gets really long, Claude's process of summarizing earlier parts uses additional tokens—and from that point on, starting a new session becomes the better choice. That said, what you experienced isn't wrong either. Usage isn't based on "number of messages" but on the tokens being processed. If you start a new session and have to re-explain the previous context at length, that explanation itself consumes tokens—which could actually cost you more. In other words, the benefit depends on the situation. Here's the bottom line: 1. If your conversation has gotten very long and you're seeing a usage limit warning, starting a new session is usually better. 2. For materials or rules you'll reference repeatedly, don't re-explain them in each new session—build them into your project setup instead. Then when you reuse them, they get cached and consume fewer tokens toward your limit (per official documentation). 3. If your conversation is still short and you'll keep needing the context, there's no need to cut it off. I've detailed this and related topics in these guides: https://usingclaude.com/ko/pricing/plans/reduce-claude-usage https://usingclaude.com/ko/guides/features/claude-long-conversation-context