DeepSeek Could Run 25 Agents at 300 tok/s Per GPU, TeortaxesTex Argues Swarms Are Coming

teortaxesTex · x · 2026-09-10

Responding to skepticism that Kimi or DeepSeek can't run large agentic swarms soon, teortaxesTex argues nothing stops it: DeepSeek could likely serve 25 V4.1 agents at 300 tok/s off a single GPU, and will soon have 100K+ GPUs.

Related event: teortaxesTex defends DeepSeek swarm scaling despite gap with Astra(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →