Verified October 2026

Home › Tools › Tokens per word: Arabic

Tokens per word in Arabic

Arabic averages about 2.9 tokens per word on current BPE tokenizers, which is 118% more tokens than the same count in English. Arabic is the most expensive major language per word: prefixes, suffixes and diacritics all split into separate tokens.

2.9Tokens per word
0.34words per token
2,900Tokens per 1,000 words
+118%vs English

Arabic to tokens table

Words → Tokens (Arabic, 2.9 tokens per word)
WordsTokens
100290
250725
5001,450
1,0002,900
2,5007,250
5,00014,500
10,00029,000
25,00072,500
50,000145,000
100,000290,000

Language comparison

LanguageTokens per wordTokens for 1,000Cost multiplier vs English
English1.331,3301.00×
Spanish1.71,7001.28×
French1.81,8001.35×
German1.91,9001.43×
Portuguese1.71,7001.28×
Russian2.42,4001.80×
Arabic2.92,9002.18×
Hindi3.23,2002.41×
Chinese1.51,5001.13×
Japanese1.81,8001.35×

Ratios are per word except Chinese and Japanese, which are per character. Comparing them directly is approximate — the honest way to budget is to tokenize a sample of your own corpus.

Cost of 1,000 words in Arabic

Cost of 2,900 tokens (1,000 words) — USD, rates verified October 2026
#ModelProviderInputOutputBlended
1GPT-6 LunaOpenAI$0.00029$0.00145$0.00065
2GPT-5.6 LunaOpenAI$0.00058$0.00348$0.00145
3GPT-5.4 nanoOpenAI$0.00058$0.00362$0.00149
4Gemini 3.5 Flash-LiteGoogle Gemini$0.00087$0.00725$0.00268
5DeepSeek V4.1 FlashDeepSeek$0.00087$0.00348$0.00174
6GPT-5.4 miniOpenAI$0.00217$0.013$0.00544
7Gemini 3.8 FlashGoogle Gemini$0.00217$0.0109$0.00489
8Gemini 3.7 FlashGoogle Gemini$0.00217$0.0109$0.00489
9Claude Haiku 4.5Anthropic$0.0029$0.0145$0.00652
10Gemini 2.5 ProGoogle Gemini$0.00362$0.029$0.0109
11Grok 4.3xAI Grok$0.00362$0.00725$0.00544
12DeepSeek V4 ProDeepSeek$0.00383$0.0115$0.0067
13Gemini 3.5 FlashGoogle Gemini$0.00435$0.0261$0.0109
14GPT-5.2OpenAI$0.00507$0.0406$0.0152
15GPT-6.1 SolOpenAI$0.0058$0.029$0.013
16GPT-6 SolOpenAI$0.0058$0.029$0.013
17GPT-5.6 TerraOpenAI$0.0058$0.0348$0.0145
18GPT-4.1OpenAI$0.0058$0.0232$0.0116
19Claude Sonnet 5Anthropic$0.0058$0.029$0.013
20Gemini 3.1 Pro (Preview)Google Gemini$0.0058$0.0348$0.0145
21Grok 4.6xAI Grok$0.0058$0.0174$0.0101
22GPT-5.4OpenAI$0.00725$0.0435$0.0181
23Claude Sonnet 4.6Anthropic$0.0087$0.0435$0.0196
24GPT-5.6 SolOpenAI$0.0116$0.058$0.0261
25GPT-5.5OpenAI$0.0145$0.087$0.0362
26Claude Opus 5Anthropic$0.0145$0.0725$0.0326
27GPT-6 AstraOpenAI$0.029$0.145$0.0653
28Claude Fable 5.1Anthropic$0.029$0.145$0.0653

Input = 2,900 prompt tokens. Output = 2,900 generated tokens. Blended = input + 25% of output, the same weighting used across this site.

What this means for a multilingual product

If you serve Arabic and English from the same budget, Arabic consumes 2.2× the tokens for the same number of words. Either the price per token absorbs that, or the Arabic product quietly runs at a lower margin. Measuring the ratio on your own text is the only reliable way to price it.

FAQ

How many tokens is 1,000 words in Arabic?

About 2,900 tokens, using 2.9 tokens per word. That is 118% more tokens than the same count in English (1,330).

Why is Arabic more expensive per word?

Arabic is the most expensive major language per word: prefixes, suffixes and diacritics all split into separate tokens. Providers bill per token, not per word, so a language that fragments into more tokens costs more for the same meaning.

How do I get an exact count for Arabic?

Run a sample of your own text through the tokenizer of the model you will be billed on. Public ratios are averages; product names, numbers and formatting in your own data can move the result by 20% or more.

Keep going

Calculate a real monthly bill