iCreatAI News
Skip to content
Sarvam AI Blog·· 16 hr agoFeaturedAI score71

Introducing Sarvam-105B-1711

Introducing Sarvam-105B-1711

AI summary

Sarvam released Sarvam-105B-1711, an open-weight Flash-Lite checkpoint of Sarvam 105B trained on upgraded B200 infrastructure, with weights on Hugging Face in BF16 and a 65 GB NVFP4 version that runs on a single workstation GPU. Its average across eight conversational agent benchmarks rose from 42.6 to 74.5 versus the February release, and inference optimisations cut its blended price from $0.44 to $0.15 per million tokens. The company reports it matches Gemini 3.5 Flash-Lite and Gemma 4 31B at a fraction of their price, with large gains on IFBench (32.5 to 82.5), Tau2 (46.2 to 85.5) and its Indic suite (30.1 to 79.1).

Why it matters

The post gives unusually detailed benchmark, pricing and on-prem deployment figures for a low-cost open-weight model aimed at Indian-language enterprise agents.

Source: Sarvam AI Blog · sarvam.ai