AILobstersabout 1 hour ago
If this is true, the hyperscalers are toast
1 min read
Stanford research shows small language models match or beat large ones in 81.2% of tasks at 50-85% lower costs. Models like QWEN 3 and GEMMA 3 run on desktop PCs with Nvidia or Apple M4 chips. This threatens data centre investments and favours Dell and Apple.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
