BPE
Exact token counts
Uses the same byte-pair encodings as OpenAI: o200k_base, cl100k_base, p50k_base and r50k_base.
See exactly how many tokens your prompt uses before you send it to an OpenAI model. Counting runs on your own device with the official BPE encodings, so your text never leaves the browser. Pick from 59 models and the right tokenizer is selected for you.
Uses the same byte-pair encodings as OpenAI: o200k_base, cl100k_base, p50k_base and r50k_base.
GPT-5.5 back to GPT-3, o-series reasoning models, gpt-oss and embeddings.
Everything runs in your browser. Nothing you paste is uploaded or stored on a server.
Auto update recounts on every edit, or switch it off and count on demand for huge files.
Colored token blocks show how text is split; hover a block for its token ID, or view raw IDs.
Check how much of the model's context window a prompt fills and estimate cost at your price.
Simple, free tools to help you build, test, and get more done.
Explore Our Free Toolkit →