Hey Product Hunt 👋 I kept hitting the same wall: model prices vary 10–100×, but the only way to know which one was actually good enough for my task was to wire up API keys, write eval scripts, and burn a weekend. Leaderboards didn’t help - they rank models on generic benchmarks, not the work I actually do. And a tool owned by a model maker can’t neutrally tell you to use a cheaper competitor. So I built TokenHunger. Paste your task and a few cases, and it runs them across every model, scores the answers, and ranks by cost per correct answer - the cheapest model that still passes, first. No API keys, no config. You get a free cost estimate without signing in, and 5 free credits when you sign in with GitHub. It’s deliberately neutral: we don’t make or resell any model, and runs sit on separate billing - nothing has a thumb on the scale. Would love your feedback: what task would you benchmark first, and which models should we make sure are covered?

TokenHunger

Find the cheapest model that still passes your task

Find the cheapest model that still passes your task