Infrastructure

Tokens

The units of text that language models read and generate, used to measure cost and limits.

Also written as token

Messages
1,034
Members using it
153
First seen
Nov 19, 2025

What it means

The fundamental units that language models process, roughly corresponding to parts of words or whole words. Models charge by tokens consumed (input) and generated (output), making token efficiency important for cost management. System prompts, context, and responses all consume tokens from available limits.

How it's used here

Members track token consumption carefully, complaining about "token bloat" from verbose system prompts and celebrating when they can't max out prepaid token allowances. The group is very cost-conscious about per-token pricing versus subscription models.

Seen in the wild

  • “I was frustrated with Claude Code's token bloat with their system prompt a few weeks back 23k tokens for every session, every subagent, ugh”
  • “tested using gemini sun but i just have one of the base things that comes with google suite setup and there's basically zero amount of tokens”
  • “tested using gemini sun but i just have one of the base things that comes with google suite setup and there's basically zero amount of tokens”
  • “Before when I used OpenClaw with 7 $25 OpenAI business accounts and found out I got use this tokens it was hard to max out a $25 account even”

More in this category

All terms