How MiniMax M3 Was Built: Sparse Attention and Native Multimodality — Olive Song

AI Engineer · 4,333 views

An inside look at how MiniMax M3 was built, covering its sparse attention design and native multimodality, from someone on the team.

More in Models
21:12Models
Deepseek crushed everyone
Matthew Berman · 136,639 views · Oct 2026
DeepSeekopen weights
47:26Models
Introducing Genie 4
Google for Developers · 55,455 views · Oct 2026
Genie 4Google DeepMindkeynote
2:28Models
Genie 4: A new generation of world simulation
Google DeepMind · 565,621 views · Oct 2026
Genie 4Google DeepMindlaunch