Lab / Tokenizer Playground
🔤

Tokenizer Playground

Visualize how LLM tokenizers break text into tokens. See BPE merges step-by-step and compare token counts across models.

Input Text
Tokens40
Byte Pair Encoding with merge rules
Hello,·world!·This·is·a·simple·test·of·the·tokenizer.
Estimated Token Counts by Model
GPT-4o
14
~4.0 c/t
GPT-4
15
~3.7 c/t
Claude 3.5
16
~3.5 c/t
Llama 3
14
~3.8 c/t
Statistics
Characters
53
Words
10
Lines
1
Tokens
40
Chars / Token
1.32
API Pricing Reference
ModelEst. TokensInput $/1MOutput $/1M
GPT-4o14$2.50$10.00
GPT-415$30.00$60.00
Claude 3.516$3.00$15.00
Llama 314FreeFree

Uses a simplified BPE implementation for demonstration. Token counts for specific models are heuristic estimates. All processing runs entirely in your browser — nothing is sent to a server.