Paste text or upload a file and see exactly how many tokens you'll use across every major LLM — plus the dollar cost per call. 100% client-side, your text never leaves your browser.
tiktoken for OpenAI, Anthropic's tokenizer for Claude). Use this for budget planning — differences are typically within ±5%. Pricing snapshot 2026-04-24 (confirm at provider docs).
Tokens are the units large language models charge for. Roughly, one token equals 3-4 characters of English text or about 0.75 of a word, but the exact ratio differs across tokenizers. OpenAI's cl100k_base and o200k_base tokenizers, Anthropic's BPE for Claude, Google's SentencePiece for Gemini, and Meta's tokenizer for Llama all yield slightly different counts on the same text.
This tool uses calibrated character-to-token ratios for each model family: roughly 3.8 chars/token for Claude, 4.0 chars/token for GPT-4o/GPT-5, 4.0 chars/token for Gemini, and 3.6 chars/token for Llama 3. Code-heavy text trends lower (more tokens per char); prose trends higher. For exact counts during production budgeting, use each provider's official tokenizer.
Cost is computed using the listed input price per million tokens, assuming the entire pasted text is the input. Output cost is shown separately at the typical 1:1 input-to-output ratio. All calculation runs in your browser — nothing is uploaded.