← back
Sovereign Escape Velocity: Ownership w Open Models — Gus Martins, & Ian Ballantyne, Google DeepMind
Takeaway
Open models offer organizations control over deployment and adaptation while allowing model size to match practical workloads.
Summary
- Google DeepMind positions Gemma as a complement to hosted Gemini for local deployment, customization, proprietary data, and infrastructure ownership.
- The presentation introduces Gemma 4 variants aimed at mobile devices and servers, including E2B, E4B, a 26B mixture-of-experts model, and a 31B dense model.
- Smaller variants support text, vision, and audio inputs with text output, while the broader argument emphasizes useful capability per unit of hardware.
- The speakers describe an Apache 2.0 licensing move and national-language deployments as ways to reduce adoption friction and support sovereign AI.
gemmaopen-modelssovereign-ai
Original description
Gemma 4's 31B model sits fourth on the LM Arena open model leaderboard. The models around it are at least twice as large; some are 20 times larger. It runs on a single GPU. Competitors at comparable quality need four or five. Ian Ballantyne and Gus Martins walk through what that size efficiency unlocks: running on a Pixel phone (the E2B and E4B models use 2B and 4B of GPU memory despite having more parameters), deploying a medical variant on two GPUs for an entire hospital, and running parallel multi agent workloads on an M4 Mac via LM Studio. The talk also covers the license shift from a custom Gemma license to Apache 2.0 — the practical effect is that sovereign institutions in Ukraine, Bulgaria, and Brazil can get legal sign off without 18 months of procurement review. Speaker info: https://x.com/gusthema / gus-martins-64ab5891 / ianballantyne https://github.com/irbg