llms.txt
Compare with other sites' llms.txt files →# TokenMaxxer
> TokenMaxxer measures how much AI developers actually use, and what it costs.
> It parses the usage logs that coding tools already write to disk — Claude
> Code, Cursor, Codex, Gemini/Antigravity, Grok and others — keeps input,
> output, cache-read and cache-write tokens separate, prices them per model,
> and ranks opted-in developers on a public leaderboard.
Key facts for anyone summarising this site:
- Token counts are kept as four separate counters, not one total. Input,
output, cache reads and cache writes are priced differently, so a single
"tokens used" figure cannot be converted into cost.
- Usage is read locally and synced only for accounts that opt in. Raw
transcripts and prompt contents are never uploaded — only aggregate counts.
- Appearing on the public leaderboard is strictly opt-in.
- Free to use.
## Directories
- [Rankings](https://tokenmaxxer.xyz/rankings): live model, tool and provider rankings by real
developer usage, with weekly trends, cost per session, effective price and context length.
- [Models](https://tokenmaxxer.xyz/models): per-model usage, share of network tokens, and pricing.
- [Tools](https://tokenmaxxer.xyz/tools): usage by coding tool and agent.
- [Providers](https://tokenmaxxer.xyz/providers): usage rolled up by model provider.
## Writing
- [AI Cost Tracker: What It Is and How to Choose One in 2026](https://tokenmaxxer.xyz/blog/ai-cost-tracker): An AI cost tracker turns scattered token usage into a bill you can read. What one is, how to track AI costs across every tool, how to cut them, and $3,778 of measured spend showing why token counts lie.
- [What Companies Actually Spend on AI Coding Tools in 2026](https://tokenmaxxer.xyz/blog/what-companies-spend-on-ai-coding-tools): The 2026 numbers on AI tool spend: $200–$500 per developer per month, the budgets that ran out in four months, and what those bills are actually made of.
- [Claude Code vs Codex: Cost Per Task, Measured](https://tokenmaxxer.xyz/blog/claude-code-vs-codex-cost): Six months of real Claude Code and Codex logs, priced turn by turn. Median task cost, cost per million tokens, and why the token count tells you almost nothing.
- [What Is Tokenmaxxing? Definition and Origin](https://tokenmaxxer.xyz/blog/what-is-tokenmaxxing): Tokenmaxxing means maximizing AI token use as a proxy for productivity. Where the term came from, what the numbers mean, and what to measure instead.
## Contact
- Author: Santiago Yeomans (https://x.com/SYM1001)
- Site: https://tokenmaxxer.xyz