Token calculator for AI models
Paste your text and see how many tokens it will consume in ChatGPT, Claude, GPT-5, Gemini, and other models. Calculated locally in your browser — your text never leaves your device.
Characters
0
Words
0
Tokens (estimated)
0
Estimated cost per model
Standard-tier API list prices per 1M tokens, as of mid-2026 — always verify with each provider, they change often. Token counts are an estimate (±5%).
What is a token?
A token is the minimum unit AI models process. In English, one token equals roughly 4 characters or 0.75 words. Models charge separately for input and output tokens, with different rates for each.
Why calculate tokens?
To estimate costs before processing long documents, fit prompts within the model's context window, and understand why the same text can cost more or less depending on the tool.
How many tokens is a document?
A rough guide — about 1.3 tokens per word in English (a little more in Spanish and other Romance languages).
Worked example: a 100-page report is roughly 100k tokens in and 100k out. Via API that's about $1.75 on GPT-5.6 (Terra) or ~$0.60 on Claude Haiku 4.5 — cheap per run, but the real cost is the hours spent stitching sections together for coherence, which is exactly what a per-document tool removes.
Need to generate a 100+ page document?
General-purpose models lose coherence on long documents. Nomos uses multi-agent architecture to maintain coherence across chapters, with predictable per-credit pricing (not per-token).
Frequently Asked Questions
Is this calculator exact?
The calculation is an approximation. Each model uses its own tokenizer (BPE in GPT, SentencePiece in Gemini). The estimate has a margin of ±5%. For mission-critical budgets, consult the official tokenizer of the model you'll use.
Is my text sent to any server?
No. All calculation happens in your browser with JavaScript. Your text never leaves your device. This page even works offline once loaded.
Why does Claude charge differently than GPT?
Each provider sets pricing based on inference cost and commercial strategy. Prices shown here are standard-tier API list prices as of mid-2026 (verify with each provider before quoting). The most capable models (Opus, GPT-5) cost more; the fast ones (Haiku, Flash, Mini) cost less.
How many tokens fit in a model?
It depends on the model, and by 2026 the windows are huge: GPT-5.6 and Claude (Opus 4.8 / Sonnet 4.6) handle around 1M tokens, and Gemini 3.x goes up to 2M. But a big window doesn't guarantee coherence — real performance still degrades as it fills up, because models weight information in the middle of a long context less (the 'lost in the middle' effect).
You might also like
ChatGPT alternative for long documents
Why ChatGPT fails on long documents and what to use instead
Free bid document analyzer
Extract criteria, budget and solvency from any tender document
AI Thesis Generator
Generate your 80-200 page thesis with chapter coherence
AI Annual Report Generator
Generate corporate annual reports with AI