About
AltusJ AI Token Tracker
A neutral, always-up-to-date dashboard for comparing the price and performance of large language models.
Why we built it
Picking the right LLM for a project is harder than it should be. Pricing pages live in different places, change without warning, and rarely sit next to the benchmark numbers that actually matter. Engineers end up with stale spreadsheets and gut-feel comparisons.
AltusJ AI Token Tracker brings the two together: live pricing on one axis, accepted public benchmark scores (MMLU, HumanEval) on the other, plus a calculator that turns those numbers into your real monthly bill.
What you get
- 100+ models across OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, xAI, Qwen, Cohere and more.
- Live input/output token prices sourced from the OpenRouter aggregator and refreshed continuously.
- Price-vs-performance scatter chart on a log price scale so cheap and expensive models stay legible together.
- Monthly cost calculator that ranks every model against your actual workload (input tokens, output tokens, calls per month).
- Custom model entries so internal or upcoming models can sit alongside the public ones for a fair comparison.
How the data is sourced
Pricing is fetched from public model catalogs (primarily OpenRouter). Benchmark figures are public estimates published by the model providers or the benchmark maintainers and are intended for relative comparison only — they are not measured by us. Where a model has not published a score, it is omitted from that chart rather than guessed.
Independence
The site is not affiliated with any model provider. We do not take sponsorship from labs whose models appear on the page. The goal is an honest signal, not a referral catalog.
Contact
Questions, corrections, or model requests? Reach out via the email listed in our Privacy Policy.