Free AI Tokens Counter
Paste or type any text and instantly see the exact token count for OpenAI models (GPT-4o, GPT-4.1, GPT-5, o1, o3, o4-mini) or get a close estimate for Claude, Gemini, and Llama. This free token counter uses gpt-tokenizer, a production-grade BPE tokenizer that runs entirely in your browser — your text never leaves your device. Use it to check prompt size before API calls, estimate costs, optimize prompts to fit context windows, and debug tokenization by viewing individual token IDs and decoded strings.
Need expert help with AI?
Looking for a specialist to help integrate, optimize, or consult on AI systems? Book a one-on-one technical consultation with an experienced AI consultant to get tailored advice.
What Is gpt-tokenizer and Why Token Counting Matters
This free AI token counter is powered by gpt-tokenizer, the fastest and most lightweight GPT tokenizer for JavaScript. It is a production-grade TypeScript port of OpenAI's tiktoken library, widely adopted in enterprise applications and open-source projects. It supports every OpenAI encoding and model — including GPT-4o, GPT-4.1, o1, o3, o4-mini, and the latest GPT-5 — giving you exact token counts that match what the API actually charges.
Token counting is essential when working with AI APIs. Every API call to ChatGPT, Claude, or other LLMs is billed by tokens — not words or characters. A single word can be one token or several, depending on the model's encoding. This tool lets you check exactly how many tokens your prompt uses before you send it, helping you stay within context window limits and estimate API costs accurately.
The tokenizer supports all OpenAI encodings (cl100k_base, o200k_base, o200k_harmony, r50k_base, p50k_base, and p50k_edit), so whether you are building prompts for GPT-3.5, GPT-4, or the newest models, the count will match what the API actually charges you. It runs entirely in your browser with no data sent to any server.
How Token Counting Works Under the Hood
If you need tokenization in JavaScript or TypeScript, gpt-tokenizer is the leading alternative to running tiktoken via WASM or Python bindings. Available on npm, it works in any JS runtime — browser, Node.js, Deno, and Bun. It is the fastest tokenizer on npm, outperforming even WASM and native binding implementations, with the smallest bundle size thanks to compact encoding storage and tree-shakeable per-model imports.
The API includes encode, decode, countTokens, isWithinTokenLimit, encodeChat for chat-format messages, and estimateCost with built-in model pricing data. It supports streaming tokenization via generators, synchronous loading, and an LRU merge cache for repeated tokenization. There are no global caches, so no memory leaks — making it production-safe for long-running server applications.
How to Count Tokens for ChatGPT, Claude, and Other AI Models
Whether you are building a chatbot, writing system prompts, or managing RAG pipelines, knowing how many tokens your text consumes is critical. Every AI API — OpenAI, Anthropic, Google, Mistral — bills by tokens, and every model has a context window limit measured in tokens. Sending a prompt that exceeds the limit causes errors or silent truncation; underestimating token usage leads to surprise costs on your monthly bill.
This tool gives you the exact count for any OpenAI model by using the same BPE encoding the API uses internally. For Claude, Gemini, Llama, and other models, the count is a reliable estimate because modern tokenizers share similar vocabulary sizes and merge strategies. Paste your prompt, system message, few-shot examples, or even a full document and see the token count in real time — no API key, no rate limits, and no data leaving your browser.
More Free Tools
More than 20 free AI tools.
Latest Articles
My take on what's new in AI and programming.
Q&A SESSION
Got a quick technical question?
Skip the back-and-forth. Get a direct answer from an experienced engineer.
How It Works
Paste or type your text into the input area.
The BPE tokenizer (gpt-tokenizer) analyzes the text locally in your browser.
View the exact token count instantly, plus a visual preview of each token and its ID.
Key Features
Privacy & Trust
Use Cases
Frequently Asked Questions
Is this AI token counter free to use?
Yes, it is completely free with no usage limits, no signup, and no API key required. The tokenizer runs entirely in your browser using the open-source gpt-tokenizer library, so there are no server costs to pass along. You can count tokens for as many prompts as you need without restrictions.
Is my text sent to a server when I count tokens?
No. All tokenization runs locally inside your browser — your text never leaves your device. There are no API calls, no logging, and no server-side processing. This makes it safe to check token counts for proprietary prompts, confidential documents, system prompts with trade secrets, or any text you would not want exposed to a third party. You can confirm this by watching the Network tab in DevTools.
Which AI models does this token counter support?
The counter uses the gpt-tokenizer library, which supports all OpenAI encodings: o200k_base (GPT-4o, GPT-4.1, GPT-5, o1, o3, o4-mini), cl100k_base (GPT-4, GPT-3.5-turbo, text-embedding-ada-002), r50k_base (GPT-3, text-davinci-003), p50k_base, and p50k_edit. The token counts will be exact for any model that uses these encodings. For non-OpenAI models like Claude, Gemini, or Llama, the counts serve as a close approximation since most modern tokenizers produce similar results for English text.
Is the token count from this tool exact or an estimate?
For OpenAI models, the count is exact — this tool uses the same tokenization algorithm (BPE with the same merge rules) that the OpenAI API uses internally. The gpt-tokenizer library is a faithful TypeScript port of OpenAI's tiktoken. For other providers like Anthropic Claude or Google Gemini, which use their own tokenizers, the count is a close approximation — typically within 5-10% of the actual value.
Can I use this to estimate API costs before making a call?
Yes, this is one of the primary use cases. By knowing the exact token count of your prompt before sending it, you can estimate the cost by multiplying tokens by the model's per-token price. For example, if your prompt is 2,000 tokens and GPT-4o charges $2.50 per million input tokens, that prompt costs about $0.005. The tool does not calculate prices directly because API pricing changes frequently, but the token count it gives you is the number you need for that calculation.
How can I use this to check if my prompt fits a model's context window?
Paste your full prompt — including system message, user message, and any examples — and compare the token count against your model's context limit. GPT-4o supports 128K tokens, GPT-4.1 supports 1M tokens, Claude supports 200K tokens, and Gemini supports up to 1M tokens. Remember that the context window must hold both your input tokens and the model's output tokens, so leave headroom. If your prompt is 100K tokens on a 128K model, you only have 28K tokens for the response.
Does this token counter work offline?
Yes. Once the page loads and the tokenizer library is initialized, token counting works entirely offline in your browser. This is useful for developers working in air-gapped environments or anyone who wants to check token counts without an active internet connection.
What is the difference between gpt-tokenizer and tiktoken?
tiktoken is OpenAI's official tokenizer written in Python and Rust. gpt-tokenizer is a community-maintained TypeScript port that implements the exact same Byte Pair Encoding (BPE) algorithm and uses the same merge rules, so both produce identical token counts for any given input. The difference is that gpt-tokenizer runs natively in JavaScript environments — browser, Node.js, Deno, and Bun — without needing Python or WASM bindings. This tool uses gpt-tokenizer so it can run directly in your browser with no server required.
What is Byte Pair Encoding (BPE) and how does it relate to tokens?
Byte Pair Encoding is the algorithm that GPT models use to split text into tokens. It starts with individual bytes and iteratively merges the most frequent pairs into single tokens, building a vocabulary of sub-word units. Common English words like "the" or "hello" become single tokens, while rare words, technical terms, code, and non-English text get split into multiple tokens. For example, "tokenization" becomes "token" + "ization" (2 tokens). This is why token count does not equal word count — it depends on how the BPE vocabulary was trained.
What is the difference between tokens and words?
A rough rule of thumb is that one token equals about 0.75 words in English, or conversely one word averages about 1.3 tokens. However, this ratio varies significantly depending on the content. Simple English prose is close to that average, but code, technical jargon, URLs, JSON, and non-English languages can use 2-4x more tokens per word. This is why counting tokens directly — rather than estimating from word count — matters when working with API limits and billing.
Can I use this to count tokens for Claude or Gemini?
Yes, with a caveat. Claude (Anthropic) and Gemini (Google) use their own proprietary tokenizers, so the counts from this tool are close approximations rather than exact matches. In practice, most modern BPE tokenizers produce similar token counts for standard English text — typically within 5-15% of each other. For precise Claude token counts, Anthropic provides a token counting API endpoint; for Gemini, Google offers countTokens in their SDK. This tool is still useful for quick estimates across all providers without needing separate API keys.
Is this tool affiliated with OpenAI?
No. This is an independent tool built on the open-source gpt-tokenizer library, which is a community-maintained TypeScript port of OpenAI's tiktoken. It is not endorsed by, affiliated with, or maintained by OpenAI. The tokenization logic is faithful to the original, which is why the counts match, but there is no business relationship involved.
Limitations
- Token counts are estimates and may vary by model
- Does not account for system or hidden prompts
- Not a billing or pricing guarantee
- Does not calculate pricing or API costs