Tool
ChatGPT Token Counter
Paste your text below to see how many tokens it consumes in ChatGPT / Claude / Gemini and how much it would cost on the API. Everything processed in your browser — nothing sent to any server.
Estimated tokens
0
≈ tokens (rule of thumb: 1 token ≈ 4 chars in English)
Words
0
Characters
0
Estimated input cost on API
- GPT-4o ($2.50/1M tokens) $0.000000
- GPT-4o-mini ($0.15/1M) $0.000000
- Claude Sonnet 4 ($3.00/1M) $0.000000
- Gemini 2.5 Pro ($1.25/1M) $0.000000
Input prices as of July 2026. Output costs 3-5x more. Prices change — verify with source.
What is a token?
AI models like ChatGPT, Claude and Gemini don't "read" letters — they read tokens, which are fragments of words. In English, 1 token ≈ 4 characters (or ¾ of a word).
Why does this matter?
If you use the API of any model (not the website), you pay per token — both input and output. A 1000-token prompt on GPT-4o costs $0.0025 of input. Seems small, but at 10,000 requests/day, that's $25/day on input alone.
How to reduce tokens
- Cut repeated instructions — the system already "remembers" the system prompt.
- Use bullets instead of long paragraphs.
- Ask for compact JSON, not formatted Markdown.
- Remove redundant examples — 2 good ones beat 10 mediocre.
- For translation, use smaller models (GPT-4o-mini is 15x cheaper).
This tool uses an estimate (not the official OpenAI/Anthropic tokenizer).
Average error: ±10% on common texts. For exact values, use OpenAI's tiktoken.