Token Counter & LLM Cost Calculator
Paste any text to count tokens and compare what it costs across 11 AI models — GPT-5.5, Claude, Gemini, Grok, DeepSeek and more. Exact token counts via OpenAI's o200k tokenizer, all in your browser. Nothing you paste leaves your device.
The answer the model writes back — usually the bigger cost.
Project recurring cost — e.g. 100K calls a month.
Cost across models
Gemini 2.5 Flash Google | $0.1000 | $0.4000 | $0.000200 | 0.0% |
DeepSeek V3.2 DeepSeek | $0.2800 | $0.4200 | $0.000210 | 0.0% |
Llama 4 Maverick (hosted) Meta | $0.2000 | $0.6000 | $0.000300 | 0.0% |
GPT-5.4 nano OpenAI | $0.2000 | $1.25 | $0.000625 | 0.0% |
Mistral Large 3 Mistral | $0.5000 | $1.50 | $0.000750 | 0.0% |
Grok 4.3 xAI | $1.25 | $2.50 | $0.001250 | 0.0% |
Claude Haiku 4.5 Anthropic | $1.00 | $5.00 | $0.002500 | 0.0% |
Gemini 2.5 Pro Google | $1.25 | $10.00 | $0.005000 | 0.0% |
Claude Sonnet 4.6 Anthropic | $3.00 | $15.00 | $0.007500 | 0.0% |
Claude Opus 4.8 Anthropic | $5.00 | $25.00 | $0.0125 | 0.0% |
GPT-5.5 OpenAI | $5.00 | $30.00 | $0.0150 | 0.0% |
Token counts use OpenAI's o200k tokenizer (GPT-4o/GPT-5); other models tokenize within ~10–15% for English text, so non-OpenAI costs are estimates. Prices last reviewed June 2026 — always confirm on the provider's page: OpenAI · Anthropic · Google · xAI · DeepSeek · Meta · Mistral.
Spend less per prompt
- Tighten the prompt itself with the prompt builder and quality score.
- Read how to cut AI prompt costs and build a model cost-routing strategy.
- Picking a budget model? Which AI model for cost-sensitive workloads, or compare specs and pricing across every model on the AI model comparison.
Frequently asked questions
How does the token counter work?
Tokens are counted with OpenAI's o200k_base tokenizer — the one GPT-4o and GPT-5 use — running entirely in your browser, so your text is never uploaded. A token is roughly 4 characters or about 0.75 words of English, but the exact count depends on the actual text (punctuation, code, and non-English text tokenize differently).
Are token counts the same for Claude and Gemini?
Not exactly. Each provider uses its own tokenizer, so Claude and Gemini counts typically land within about 10–15% of the OpenAI count shown here for English prose. The cost figures use the OpenAI token count as a common basis, so treat non-OpenAI costs as close estimates rather than exact invoices.
How is the cost calculated?
Cost = (input tokens ÷ 1,000,000 × input price) + (expected output tokens ÷ 1,000,000 × output price), multiplied by the number of calls. Input is your pasted text; you set the expected output length and call volume. Every provider prices per million tokens, with output usually costing several times more than input.
Why is output more expensive than input?
Generating tokens is more compute-intensive than reading them, so most providers charge 3–5× more per output token than per input token. That is why a short prompt that produces a long answer can cost more than a long prompt with a short answer — and why trimming requested output length is often the biggest cost lever.
Is the pricing up to date?
Prices are reviewed periodically and each model links to its provider pricing page, but LLM pricing changes often. Always confirm against the official source before budgeting. The token counts are exact; the dollar figures are best-effort estimates.
Does my text get sent anywhere?
No. Tokenization and all cost math run client-side in your browser. Nothing you paste is sent to SurePrompts or any server.