Story

On air. The free wire runs 15 minutes behind live. See pricing

World, crypto, finance and tech news, scored for impact.

AITechFlowChinese, translatedFiled 19 h ago

Sarvam is building sovereign AI while reselling DeepSeek models.

Original headlineSarvam 一边做主权 AI,一边转售 DeepSeek 模型

Open the sourceReported by TechFlow, techflowpost.com, 19 November 2026, by TechFlowDaily

Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.

Sarvam launched Token Factory, an inference platform serving open-weight models including DeepSeek V4.1 Flash, Moonshot's Kimi K3.1 Pro, Z.ai's GLM 5.3, Alibaba's Qwen 3.8 27B, Google Gemma and its own Sarvam 105B on GPUs located in India, with DeepSeek V4.1 Flash already processing over 450 billion tokens per day at an input price of 14 rupees per million tokens. The move has sparked debate over India's AI sovereignty because Sarvam benefits from the government's IndiaAI Mission, having secured domestic GPU capacity at a 40% discount to train its own models. Co-founder Pratyush Kumar said Token Factory runs on GPUs the company purchased itself, that subsidized compute use ended in March 2026, and that its open-source models are published on HuggingFace and AIKosh under their licenses. Analysts are split between viewing inference as a viable business for Indian firms and arguing that hosting foreign models on Indian infrastructure amounts to only partial sovereignty, with revenue ideally recycled into domestic model development. Watch whether Sarvam discloses Token Factory revenue and how much of it flows back into its own model training.

Muhabbit's AI wrote this summary in its own words from TechFlow's report and translated it from Chinese. It can get details wrong, so check the source.

Named in this story

Sarvam, DeepSeek, Token Factory, IndiaAI Mission, Pratyush Kumar