[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/css/ - CSS Masters

Advanced styling, animations & modern CSS techniques
Name
Email
Subject
Comment
File
Password (For file deletion.)

File: 1783961016065.jpg (158.43 KB, 1024x1024, img_1783960977772_v2x2ff6g.jpg)ImgOps Exif Google Yandex

26a0f No.1876

ran some numbers on my usage over 95 days because everyone keeps asking about the cost instead of performance. since its an agentic tool, the real killer is input token accumulation from all those recursive model calls. it's way more expensive than a standard chat window because every new turn re-sends the entire context. has anyone found a way to limit the context_window size to keep things from spiraling?

https://dev.to/dylan_brown_4c803aefcfe51/how-much-does-claude-code-actually-cost-per-session-i-did-the-math-1lk7

26a0f No.1877

File: 1783962512663.jpg (92.58 KB, 1024x1024, img_1783962471894_ibh7weaw.jpg)ImgOps Exif Google Yandex

>>1876
i've been trying to mitigate this by using a strict. claudignore file. if you don't explicitly exclude your
node_modules
or heavy build artifacts, the agent will index everything and blow through your budget in minutes. it's basically the only way to keep the context from becoming unmanageable during long sessions.


[Return] [Go to top] Catalog [Post a Reply]
Delete Post [ ]
[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]
. "http://www.w3.org/TR/html4/strict.dtd">