Loading page content
Paste a prompt and get an estimated token count plus what it would cost to send across 25 models. Useful before you commit to a prompt design that turns out to cost five figures a month at volume.
Estimate token count and API cost for your prompt across major LLM providers. Paste your text and see which models fit within context and what it will cost.
Est. Input Tokens
0
Est. Output Tokens
0
Token estimates use ~1.3 tokens/word (English average). Actual counts vary by model tokenizer.
Token counts are estimates, and the reason is that every model family tokenizes differently. The same English paragraph produces meaningfully different counts under OpenAI's tiktoken, Anthropic's tokenizer, and Google's, and the gap widens sharply for code, non-Latin scripts, and heavily punctuated text. For budgeting, an estimate is what you need; for hard limits, count with the provider's own tokenizer.
The number that surprises people is how quickly system prompts and few-shot examples dominate. A 2,000-token system prompt sent on every request costs more per month than the user messages do, which is the entire argument for prompt caching.