Open model · Exact tokenizer
DeepSeek R1 Token Counter
Load the DeepSeek R1 tokenizer from Hugging Face and inspect how your prompt is split into tokens and token IDs.
Accuracy note
Exact base-text tokenizer output from the published Hugging Face tokenizer. Serving templates and provider-specific message wrappers can add tokens.
Frequently asked questions
Does this use the DeepSeek R1 tokenizer?
Yes. The browser downloads the published Hugging Face tokenizer on first use.
Why might an API report a different count?
An inference provider can add a chat template, special tokens, or other request formatting around your text.