Cost Calculator
Estimate what your workload would cost across every model on the leaderboard. Set your monthly token volume once, and compare every provider's published price side by side.
Prices updated 2026-07-12| Model | Input $/1M | Output $/1M | Est. monthly cost |
|---|
Each provider's published list price per 1M tokens. Open-weight models (marked ~) show a representative hosted rate that varies by provider and can also be self-hosted. Figures exclude prompt-caching and batch discounts.
Questions
How is the monthly cost calculated?
Monthly cost = input millions × input price + output millions × output price, where prices are each provider's published pay-as-you-go API rate in USD per million tokens. So 20M input tokens on a $2/1M model is $40. Rates refresh alongside the weekly leaderboard update.
What's a token, roughly?
A token is a chunk of text, around 4 characters or three-quarters of an English word on average. A page of text is roughly 500 tokens; a typical chatbot reply is a few hundred. Tokenizers differ slightly between models, so treat estimates as ballpark rather than exact.
Why do open models show a "~" on their prices?
Open-weight models like Llama, Qwen and Gemma are served by many competing hosts at different rates, or self-hosted for free plus hardware. The prices here are representative hosted rates, so your provider may charge more or less.