Feed
Sarvam AI Launches Token Factory for Local LLM Inference
AI

Sarvam AI Launches Token Factory for Local LLM Inference

17 Nov 2026, 3:20 pm · 3d ago · 1 min read · Inc42

Indian AI unicorn Sarvam AI has launched Token Factory, offering developers and enterprises access to proprietary and open-weight global artificial intelligence models through an OpenAI-compatible API. Hosted entirely on India-based GPU clusters, the platform features Sarvam 105B alongside Google’s Gemma 4 31B, DeepSeek V4.1 Flash, GLM 5.3, and Kimi K3.1 Pro. Sarvam prices its 105B model at ₹15 per million input tokens and ₹60 per million output tokens. The company claims the platform delivers 330 tokens per second with 200-millisecond latency, aiming to provide high-speed, cost-effective inference hosting within India's sovereign data ecosystem.