Cost-per-1K tokens converter
Provider A quotes “$2.50 per million tokens,” Provider B says “$0.003 per 1K,” and Provider C charges “a flat $0.01 per request.” Which is cheapest? This tool normalizes any of those to a single unit — cost per 1,000 tokens — so you can compare them directly without spreadsheet juggling.
How it works
Pick the unit your raw price is quoted in. Per 1M tokens divides by 1,000 to reach per-1K. Per 1K tokens passes straight through. Per request needs the tokens-per-request figure: it divides the flat charge by that token count, then scales to 1,000. The result is shown both per-1K and back in per-1M so it slots into whichever mental model you use, and you can normalize input and output prices separately since they almost always differ.
Tips and notes
Always normalize input and output independently — a model with a cheap input price but expensive output can lose to a flatter-priced competitor on response-heavy workloads. Once everything is in the same unit, multiply by your expected tokens-per-call to get a true per-call comparison. For a ready-made table of normalized model prices, see the model pricing comparison tool.