金色财经|Nov 19, 2026 05:01
[Sarvam Launches Inference Platform, Reselling Chinese Open-Source Models on Indian GPUs] According to a Bloomberg report on November 19th, Indian AI company Sarvam opened its inference platform, Token Factory, to developers and enterprises this week. The platform hosts open-weighted models from DeepSeek, Kimi, Z.ai GLM, Alibaba Qwen, and its own Sarvam model on GPUs within India, priced in rupees. The company claims that DeepSeek V4.1 Flash processes over 450 billion tokens daily, with input prices starting at 14 rupees per million tokens. Sarvam is a beneficiary of the Indian government's IndiaAI Mission and previously obtained domestic GPUs at a 40% discount to train its own models, sparking debate about whether it's self-developed or resold. Co-founder Pratyush Kumar stated that the platform uses GPUs purchased by the company, and the subsidized computing power usage ended in March 2026. The open-source models generated from the training have been released under license on HuggingFace and AIKosh. Analysts are divided: some believe that inference is a viable business for Indian companies, while others argue that hosting foreign models on Indian infrastructure is only "partial sovereignty" and that the revenue should be used to support domestic model development.
Share To
Timeline
HotFlash
APP
X
Telegram
CopyLink