Sarvam AI launches Token Factory serving stack with Kimi and GLM on B200s

itsOmSarraf_ · x · 2026-11-17

Sarvam AI's Token Factory is live. An engineer describes building the serving stack end to end: hosting models like Kimi K3.1 Pro and GLM 5.3 on B200 clusters, with multi-tier KV caching and prefill/decode handoff. The thread promises more inference engineering details.

Related event: India's Sarvam Launches Token Factory, Hosting Open-Weight Models Domestically(15 posts)→

Original post →

More from Infra

Infra channel →