Token Counter
Count words, characters, sentences, paragraphs and estimate AI model token counts & API costs instantly.
🧩 Estimated Token Counts
GPT-4o / GPT-4
≈ 0
cl100k_base tokenizer
GPT-3.5 Turbo
≈ 0
cl100k_base tokenizer
Claude (Sonnet / Opus)
≈ 0
Similar to cl100k_base
Gemini
≈ 0
SentencePiece-based
Llama 3
≈ 0
SentencePiece tokenizer
Mistral
≈ 0
SentencePiece-based
💰 Estimated API Cost (Input)
GPT-4o
$2.50 / 1M tokens
≈ $0.00
Claude Sonnet
$3.00 / 1M tokens
≈ $0.00
GPT-4
$30.00 / 1M tokens
≈ $0.00
Gemini Pro
$1.25 / 1M tokens
≈ $0.00
How does token estimation work?
This tool uses a heuristic algorithm to estimate token counts. It analyzes the ratio of ASCII characters, CJK (Chinese/Japanese/Korean) characters, and other Unicode characters in your text. English text averages ~4 characters per token for GPT-4 and ~3.5 for Llama models. CJK characters average ~1.5 characters per token for GPT-4 and ~1.2 for Llama. The estimates are approximate (≈) and may differ from actual tokenizer output by 5-15%.