Due to Byte-Pair Encoding (BPE) character splits, Korean text generates 1.95× more tokens than equivalent English text. 1,000 Korean words consume ~2,599 tokens in Hy-MT2-30B-A3B (OpenRouter).
Quick answer
For 1,000 Korean words, Hy-MT2-30B-A3B (OpenRouter) is modeled at 2,599 tokens—1.95× the English baseline. That is approximately $0.000192 as input, before any cache discount.
Method & trust
The language page applies the published language multiplier to a 1,333-token English baseline, then applies the model's input/output rates. Tokenizers differ, so benchmark representative text before production budgeting.
Relative to English baseline (1.0×).
Price to send 1,000 words of Korean text into context.
Discounted price for repeated Korean system context.
| Word Count Scale | Korean Tokens | Korean Input Cost | English Equivalent Cost | Tokenization Penalty |
|---|---|---|---|---|
| 1,000 words (Short Article) | 2,599 | $0.000192 | $0.000099 | +$0.000094 |
| 10,000 words (Whitepaper / Report) | 25,994 | $0.001924 | $0.000986 | +$0.000937 |
| 50,000 words (Book / Corpus) | 129,968 | $0.009618 | $0.004932 | +$0.004686 |
| 100,000 words (Enterprise Repository) | 259,935 | $0.0192 | $0.009864 | +$0.009371 |
Can consume almost 2× the tokens of English for equivalent semantic content.
Prompt caching is particularly impactful for Korean system prompts to offset the higher per-word token weight.
Most LLM tokenizers are primarily trained on English-heavy web datasets. Non-Latin characters in Korean (Hangul) split across multiple Byte-Pair Encoding (BPE) sub-word or multi-byte UTF-8 tokens, requiring approximately 1.95× more tokens to encode the exact same semantic meaning as English.
Prompt caching is particularly impactful for Korean system prompts to offset the higher per-word token weight. In addition, enabling prompt caching on static Korean instructions or documentation saves 75–90% on input token rates.
Yes. Hy-MT2-30B-A3B (OpenRouter) has strong multilingual comprehension and generation capabilities in Korean (한국어). The difference is purely computational and financial due to sub-word token splits.