Browser utility / tokenizer comparison

Token counter

Paste a skill file, prompt, or code block. The same text is counted with each tokenizer in a separate worker, entirely in your browser.

WASM / workers

Input

0 chars

Compact tokenizer assets are served locally and loaded on first use. Your text stays in the local Web Workers.

Counts

0/5 ready

Models to compare

GPT-2

R50k · byte-level BPE

Idle

openai-community/gpt2

Phi-4

SentencePiece BPE

Idle

microsoft/phi-4

GPT-OSS

o200k · byte-level BPE

Idle

openai/gpt-oss-20b

Qwen 3

Qwen 2 · byte-level BPE

Idle

Qwen/Qwen3-8B

OLMo 2

OLMo 2 · byte-level BPE

Idle

allenai/OLMo-2-1124-7B

Supported tokenizers

GPT-2 (R50k · byte-level BPE), GPT-4 (cl100k · byte-level BPE), Phi-4 (SentencePiece BPE), GPT-OSS (o200k · byte-level BPE), Qwen 3 (Qwen 2 · byte-level BPE), Qwen 3.5 (Qwen 3.5 · byte-level BPE), OLMo 2 (OLMo 2 · byte-level BPE), DeepSeek V3 (DeepSeek V3 · byte-level BPE), Nemotron 3 (Nemotron · byte-level BPE), Kimi K2 (Kimi · tiktoken BPE), Qwen 2 (Qwen 2 · byte-level BPE), TinyLlama (Llama 2 · SentencePiece BPE), OLMo 3 (OLMo 3 · byte-level BPE), CodeLlama (Llama 2 · SentencePiece BPE), Gemma 3 (Gemma · SentencePiece BPE), Gemma 4 (Gemma · byte-level BPE), Llama 3.1 (Llama 3 · byte-level BPE), Llama 3.3 (Llama 3 · byte-level BPE), Llama 4 (Llama 4 · byte-level BPE), Phi-4 mini (Phi-4 mini · SentencePiece BPE), Mistral (Mistral · SentencePiece BPE), Hunyuan 3 (Hunyuan 3 · byte-level BPE), Gemma 2 (Gemma · SentencePiece BPE), GLM 4.7 (GLM · byte-level BPE), GLM 5.2 (GLM · byte-level BPE)