SOTA Model Board · Tools

What does your prompt actually cost?

Paste any prompt. Get a live per-model price breakdown across every frontier contender — powered by real-time OpenRouter rates. Estimates run in your browser. We store anonymous usage metadata (token counts, cost figures — never your text); your prompt text is stored only if you tick the opt-in box below. Details in the privacy policy.

Your prompt

TOKENS (EST.)0 CHARACTERS0 WORDS0

Token count is a heuristic estimate (~±15%). Exact counts vary per model tokenizer — Sonnet 5's new tokenizer, for example, runs ~30% heavier than most.

Cost per model 1 run · input + output

ModelInput $Output $Total / run× 10K runs
Connecting to price feed…

FEED: connecting…