
Data residency and the lack of inference in India have kept a lot of teams here off the best open models, despite the favourable cost-accuracy trade-offs. Today we are launching in GA hosted endpoints for our and other open models for use-cases across chat, code, co-work, and (soon) cyber. With kernel optimisation, three tier KV caching, infra resiliency, and datacenter scale gpu management, we have achieved cost and reliability metrics that match and often exceed global peers. You will find a lot of technical detail in our release blog.
Today, we are launching Sarvam Token Factory for everyone. We host our own models, and the full range of best-in-class open-weight models, in India. We match leading inference providers on performance and reliability. For enterprises, this means more models to choose from, and an AI stack they can own without lock-ins.
























