Token-Zähler
OfflineToken-Anzahl für KI-Modelle berechnen.
- Bring Your Pricing
0
Chars
0
Bytes (UTF-8)
0
Words
0
Lines
Estimates
Enter rates to compute cost| Family | Input | Output | Total |
|---|---|---|---|
| GPT family (BPE) | 0 | 0 | 0 |
| Claude family | 0 | 0 | 0 |
| Llama / Gemini family (SentencePiece) | 0 | 0 | 0 |
Cost (USD)
GPT family (BPE)
Claude family
Llama / Gemini family (SentencePiece)
GPT
$0.00
Claude
$0.00
Llama
$0.00
Why counts differ
- Each model family uses a different tokenizer (BPE / SentencePiece / custom). Rows are per tokenizer family, not per model version — Gemini tokenizes like the SentencePiece row.
- Non-ASCII text (Korean/Japanese), code, and JSON often tokenize differently than plain English.
- Chat APIs may add hidden tokens for message formatting.
How this tool estimates
We estimate tokens from UTF-8 byte length and character makeup (ASCII vs non-ASCII), then apply a small safety factor in Conservative mode.
For exact billing, rely on your provider’s official tokenizer when available.