Token counter · ChatGPTChatGPT token counter
We count tokens the way OpenAI does: the o200k vocabulary for GPT-4o and newer, cl100k for GPT-4 and GPT-3.5.
Counted in your browser · vocabularies load from bringer.ru, not from third-party servers
01
Count
pick a model vocabularyTokens
—
Characters
175
Words
29
Characters per token
—
loading the vocabulary…
Context window— of 128,000 used · —
02
Which tokenizer each model uses
Example · “привет”
o200kпривет2
cl100kпривет4
03
About OpenAI tokens
OpenAI publishes its tokenizers, so counting in the browser matches what the model sees. With GPT-4o the company moved to the o200k vocabulary, twice as large, and Russian text became noticeably cheaper in it. API requests add a few service tokens to every message — the counter does not include them.
o200k_base
Vocabulary≈200k
Used by GPT-4o, GPT-4.1, the o-series and GPT-5.
cl100k_base
Older vocabulary≈100k
Used by GPT-4, GPT-3.5 and the text-embedding-3 models.
o200k_harmony
gpt-oss+ service tokens
The vocabulary of the open gpt-oss models: the same tokens as o200k plus service ones for the chat format.
Привет
Example2 and 4
The Russian word “привет” is 2 tokens in o200k and 4 in cl100k; “программирование” is 3 and 6.
Service tokens
API≈3 per message
A chat request adds a few tokens to every message — the exact figure depends on the format.
As at OpenAI
Accuracytiktoken
The vocabularies are the same as in OpenAI's tiktoken library.
04
Frequently asked questions
Updated
Counting runs in your browser. We do not show prices: they change, and the bill also includes the answer, reasoning and service tokens.