🔢 Token 计数器（GPT 分词 / tiktoken）Token Counter (GPT Tokenizer, tiktoken)
在浏览器本地用 OpenAI 的 o200k_base（GPT-4o、GPT-4.1、o 系列）和 cl100k_base（GPT-4、GPT-3.5）分词器计算文本的 token 数，同时统计字符、英文词数和中文字数，彩色显示每个 token 的切分结果和 ID，文字不会上传。Count tokens locally with OpenAI’s o200k_base (GPT-4o, GPT-4.1, o-series) and cl100k_base (GPT-4, GPT-3.5) tokenizers, plus characters, words and Chinese characters, with each token shown in colour and its ID — your text is never uploaded.
分词器文件约 1–3 MB，首次使用时从本站加载，之后在浏览器里离线计算，文字不会上传。Claude、Gemini、Llama、Qwen、DeepSeek 等模型使用各自的分词器，结果会不同；可把这里的数字当作粗略参考（英文通常相差 10%–30%，中文差异可能更大）。The tokenizer files (about 1–3 MB) load from this site on first use, then everything runs offline in your browser; your text is never uploaded. Claude, Gemini, Llama, Qwen, DeepSeek and other models use their own tokenizers, so counts differ — treat these numbers as a rough guide (often 10%–30% apart for English, more for Chinese).
使用说明与常见问题How-to & FAQ
「Token 计数器（GPT 分词 / tiktoken）」是什么？What is the Token Counter (GPT Tokenizer, tiktoken)?
在浏览器本地用 OpenAI 的 o200k_base（GPT-4o、GPT-4.1、o 系列）和 cl100k_base（GPT-4、GPT-3.5）分词器计算文本的 token 数，同时统计字符、英文词数和中文字数，彩色显示每个 token 的切分结果和 ID，文字不会上传。Count tokens locally with OpenAI’s o200k_base (GPT-4o, GPT-4.1, o-series) and cl100k_base (GPT-4, GPT-3.5) tokenizers, plus characters, words and Chinese characters, with each token shown in colour and its ID — your text is never uploaded.
如何使用「Token 计数器（GPT 分词 / tiktoken）」？How do I use the Token Counter (GPT Tokenizer, tiktoken)?
粘贴文字或提示词，选择分词器，上方立即显示 token 数和字符数；下面用不同颜色标出每个 token，勾选“显示 token ID”可看到编号。首次使用会从本站加载 1–3 MB 的分词表。Paste text or a prompt and choose a tokenizer; the token and character counts appear at once, and each token is coloured below (tick “Show token IDs” for the numbers). The first use loads a 1–3 MB vocabulary from this site.
一个汉字是几个 token？How many tokens is a Chinese character?
取决于分词器：o200k_base 对中文更友好，常用汉字多为 1 个 token，部分词语会合并；cl100k_base 中很多汉字需要 1–2 个 token。用本工具可以直接比较两者。It depends: o200k_base handles Chinese better (common characters are usually one token, some words merge), while in cl100k_base many characters take 1–2 tokens. Compare both here.
能计算 Claude、Gemini、Qwen 的 token 吗？Can it count Claude, Gemini or Qwen tokens?
这些模型使用自己的分词器，本工具的结果只能作为参考，英文通常相差 10%–30%。精确数值请使用对应厂商的 token 计数接口或分词器。Those models use their own tokenizers, so treat these counts as an estimate (often 10%–30% apart for English); use the vendor’s tokenizer for exact numbers.
「Token 计数器（GPT 分词 / tiktoken）」免费吗？文件会上传吗？Is the Token Counter (GPT Tokenizer, tiktoken) free? Are my files uploaded?
「Token 计数器（GPT 分词 / tiktoken）」完全免费，无需注册。你输入的内容只在本机浏览器里处理（并保存在本机方便下次继续），不会上传到任何服务器，适合处理未发表的论文和数据。The Token Counter (GPT Tokenizer, tiktoken) is free with no sign-up. Everything you enter is processed in this browser only (and kept locally so you can continue later); nothing is uploaded, so it is safe for unpublished papers and data.