AI Token Counter & Cost Calculator

Paste any prompt to estimate its token count, then compare real per-call and monthly API costs across GPT-5, Claude, Gemini and DeepSeek.

🔒 100% in-browser — your text is never uploaded
0
Est. Tokens
0
Words
0
Characters
0
Sentences

Cost calculator

Auto-filled from your text above. Edit freely.
Typical reply length you expect.
Calls across your whole app.
Provider Model In $/1M Out $/1M Cost / call Cost / month

Prices per 1M tokens, standard (non-cached, non-batch) tier, verified June 2026. Token counts are estimates (±10–15%); exact counts require each model’s own tokenizer. Always confirm live rates: OpenAI, Anthropic, Google, DeepSeek. Tip: prompt caching can cut input cost up to ~90% on repeated prefixes, and batch APIs are ~50% off.

Stop picking models by hand

You just compared 15 models in a table. QADIR OS does this on every single request — routing to the cheapest model that still clears your quality bar. Opus-grade output at Haiku-grade cost, automatically.

✓ You’re on the list. We’ll be in touch.

How token counting & pricing work

How many tokens is a word?

For typical English text, one token is roughly 0.75 words, or about 4 characters. So 100 tokens is about 75 words. Code, punctuation-heavy text, and non-English languages pack more tokens per word, so treat the number above as a close estimate rather than an exact count.

How is LLM API cost calculated?

Providers bill input (prompt) tokens and output (completion) tokens separately, priced per million tokens. Cost per call = (input tokens ÷ 1,000,000 × input price) + (output tokens ÷ 1,000,000 × output price). Monthly cost multiplies that by your number of requests — which is exactly the math the table above runs live.

Which AI model is cheapest?

Among current models, DeepSeek-V3 and Gemini 2.5 Flash-Lite are lowest-cost per token, while flagship reasoning models like GPT-5.5 and Claude Opus 4.8 cost the most. The model that matters is the cheapest one that still meets your quality bar — and that changes per task, which is why automated routing beats a fixed choice.

Is my text sent to a server?

No. This tool runs entirely in your browser. Your text is never uploaded, logged, or stored anywhere.