AI / LLM Token Cost
API cost across Claude, GPT, Gemini, or custom prices.
Math
Tokens are roughly 4 characters of English, or 0.75 words. Prompt caching can drop input cost 80-90% for repeated prefixes; batch APIs are often half-price. Check your provider's current pricing page; numbers here are 2026 list prices and change frequently.
About
Pick a model (Claude, GPT, Gemini, or custom). Enter input/output token counts per call and total calls. Get cost per call and total cost. Useful for budgeting LLM-powered features.
How to use
- Pick a model.
- Enter input/output tokens.
- Enter total calls.
FAQ
How many words is a token?+
About 0.75 English words. So 1000 tokens is roughly 750 words or 4000 characters. Code and non-Latin scripts use more tokens per character.
Can I get costs lower than this?+
Yes. Prompt caching cuts repeated-prefix input cost by 80-90%. Batch APIs are typically half-price. Multi-step prompts can often be merged. Many features that look expensive at list price are 5-10x cheaper after optimization.