Count tokens and calculate LLM API cost
Paste a prompt, pick a model, and see the exact token count plus what it costs per call, per day and per month across 17 models from OpenAI, Anthropic, Google, DeepSeek and more. Free, no signup, nothing leaves your browser.
Estimated cost
USD · GPT-5No prompt yet — the figures below only price the output side, so they are a floor, not your real cost. Paste a prompt to get a full estimate.
- Per request
- $0.005
- Per day
- $5.00
- Per month (30d)
- $150.00
- Per year
- $1800.00
Cheapest options for this workload
- GPT-5 nanoOpenAI$0.0002/req
- DeepSeek V3.2 (chat)DeepSeek$0.00021/req
- GPT-4o miniOpenAI$0.0003/req
- Llama 4 MaverickMeta$0.000425/req
Token counts use the GPT-4o/5-compatible tokenizer. Claude, Gemini and Llama tokenize slightly differently — expect ±10% on those models. Prices are list prices per 1M tokens.
Cheapest models by list price
Ranked by input + output price per 1M tokens. See the full pricing table →
Frequently asked questions
›How is the token count calculated?
We run the GPT-4o/GPT-5 compatible BPE tokenizer directly in your browser. For OpenAI models this matches the API exactly. Claude, Gemini, Llama and Qwen use slightly different tokenizers, so counts for those models are typically within about 10%.
›Does my text get uploaded anywhere?
No. The tokenizer runs entirely client-side in JavaScript. There is no backend, no account, and no analytics on your input. You can even disconnect from the network after the page loads and it will keep working.
›Why is output more expensive than input?
Output tokens require sequential generation, so providers price them 2x to 8x higher than input tokens. That is why a shorter answer with a long prompt can still be cheaper than the reverse — always model both sides.
›How do I estimate my monthly bill?
Enter your average prompt length, a realistic output length, and the number of requests per day. The calculator multiplies that out to a monthly and yearly figure. For production workloads, add 15-25% headroom for retries, system prompts and conversation history.
›What are the cheapest models right now?
On pure input + output price, GPT-5 nano, DeepSeek V3.2 (chat) and GPT-4o mini sit at the bottom of the list. For reasoning-heavy work, mid-tier models like GPT-5 mini and Gemini 2.5 Flash often beat the absolute cheapest options once you count retries.