Sarvam 2B
India's first open-weights foundational Indic LLM, trained from scratch.
About
Sarvam 2B is a 2-billion-parameter language model trained from scratch by Bengaluru-based Sarvam AI, launched August 2024. Built on 4 trillion tokens with half in 10 Indic languages (Hindi, Tamil, Telugu, Bengali, Marathi, and more), it is designed as an 'Indic Llama' rather than a fine-tune of an English-first model.
Sarvam reported it outperforming larger models like Llama 3.1 8B on Indic translation tasks despite its small size, targeting India's cost-per-query constraints and edge deployments. License caveat: weights are available for non-commercial and research use; commercial use requires explicit permission from Sarvam AI.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 90; Trending score 76/100 from researcher assessment.