ResearchBenchmarksTokenization Efficiency on Sindhi (snd)Follow1.02Bytes per TokenSarvam-2B0.84482.02743.214.3926Nov 5, 2025Evaluation ResultsMethodMethodLinksBytes per TokenSarvam-2B2025.111.02Sutra2025.113.49LLaMA-42025.113.69Gemma-32025.113.87GPT-oss2025.114.46MUTANT-Indic2025.115.4