Compare Claude, GPT, Gemini and others side-by-side. Per-call and monthly projections. 100% client-side. No signup. No tracking.
Inputs
About the numbers
Prices are per 1 million tokens, in USD, sourced from public pricing pages as of early 2026. They will drift — verify against the provider before betting real money.
Cached input tokens use the cache-read price (typically ~10% of standard input). Uncached input still counts at the normal rate.
Output is almost always more expensive than input — usually 3–5×.
Token counts are model-specific in reality. This tool treats them as comparable for planning purposes; for exact accounting use the provider's tokenizer.