List your company

Tokenizer Compression on C4 en

4.64Bits per TokenQwen 3
4.6324.6864.744.794Feb 19, 2026

Evaluation results

MethodLinks
Qwen 3Vocabulary Size=152k2026.024.64
DeepSeek R1Vocabulary Size=128k2026.024.7
Llama 3Vocabulary Size=128k2026.024.73
GPT-OSSVocabulary Size=200k2026.024.79
TrinityVocabulary Size=200k2026.024.84