Polski · interactive report / TRAVNY
AI Token Worldcup: 27 languages, six tokenizers
In this sample, Chinese uses 89 GPT-5 tokens; English uses 95. Polish / English token counts: GPT-5 (o200k_base): 123 / 95; Claude Opus 5.5: 199 / 128; DeepSeek V4.1: 141 / 95.
Results
| Language | GPT-5 (o200k_base) | Claude Opus 5.5 | DeepSeek V4.1 | cl100k_base | r50k_base | o200k_harmony |
|---|---|---|---|---|---|---|
| Chinese | 89 | 117 | 86 | 121 | 213 | 89 |
| English | 95 | 128 | 95 | 95 | 83 | 95 |
| Dutch | 99 | 187 | 124 | 127 | 133 | 99 |
| Norwegian | 104 | 153 | 128 | 129 | 131 | 104 |
| Swedish | 104 | 155 | 126 | 125 | 133 | 104 |
| Spanish | 106 | 161 | 117 | 117 | 121 | 106 |
| French | 107 | 166 | 124 | 124 | 128 | 107 |
| German | 108 | 204 | 126 | 131 | 153 | 108 |
| Turkish | 108 | 197 | 152 | 144 | 163 | 108 |
| Vietnamese | 108 | 223 | 161 | 183 | 295 | 108 |
| Portuguese | 109 | 187 | 124 | 122 | 140 | 109 |
| Indonesian | 110 | 212 | 114 | 132 | 163 | 110 |
| Arabic | 111 | 212 | 126 | 205 | 297 | 111 |
| Hebrew | 112 | 207 | 132 | 279 | 321 | 112 |
| Russian | 114 | 154 | 125 | 167 | 349 | 114 |
| Latin | 118 | 189 | 128 | 129 | 128 | 118 |
| Italian | 120 | 181 | 125 | 136 | 148 | 120 |
| Finnish | 121 | 214 | 142 | 159 | 156 | 121 |
| Polish | 123 | 199 | 141 | 152 | 190 | 123 |
| Ukrainian | 130 | 170 | 157 | 201 | 370 | 130 |
| Hindi | 133 | 275 | 214 | 384 | 550 | 133 |
| Korean | 134 | 207 | 146 | 196 | 379 | 134 |
| Slovak | 135 | 182 | 149 | 160 | 190 | 135 |
| Czech | 136 | 160 | 140 | 166 | 200 | 136 |
| Japanese | 137 | 156 | 129 | 187 | 250 | 137 |
| Hungarian | 141 | 176 | 143 | 153 | 165 | 141 |
| Greek | 149 | 260 | 186 | 323 | 397 | 149 |
Method
Each language conveys the same neutral facts about balls, arithmetic and equal division. Samples are natural formulations, not word-for-word translations. GPT-5 uses o200k_base via gpt-tokenizer; DeepSeek uses its official pinned V4.1 tokenizer. Claude Opus 5.5 counts come from Anthropic count_tokens with eight framing tokens subtracted. The three additional GPT encodings are cl100k_base, r50k_base and o200k_harmony.
Limitations
One short sample per language, not a representative corpus. Counts depend on wording, script and tokenizer. They are not a general language ranking, a measure of model quality, or a prediction of total API cost.
Reproduce and cite
Download the exact corpus and token counts (JSON) · Quarto source, recount scripts and raw measurement files · Interactive report and charts.