List your company

Tokenizer compression on C4 ko

1.72C/T RatioGPT-OSS
1.24161.36581.491.6142Feb 19, 2026

Evaluation results

MethodLinks
GPT-OSSVocabulary Size=200k2026.021.72
Llama 3Vocabulary Size=128k2026.021.68
DeepSeek R1Vocabulary Size=128k2026.021.47
Qwen 3Vocabulary Size=152k2026.021.47
TrinityVocabulary Size=200k2026.021.26