Side-by-side comparison of Gemma 4 31B Instruct (Google DeepMind · United States) and MiniMax M2 (MiniMax · China) for self-hosted deployment of the open-weight model. Gemma 4 31B Instruct is rated conditional; MiniMax M2 is conditional. They part ways on licence: Gemma 4 31B Instruct is "Apache 2.0", MiniMax M2 is "Modified MIT".
| Field | ||
|---|---|---|
| Summary | ||
| Verdict | Conditional Based on published licence terms, Gemma 4 31B ships under pure Apache 2.0 — a notable break from the Gemma Terms of Use used in prior generations — with no prohibited-use carve-outs. US origin carries Schrems-II and CLOUD-Act exposure, and at 30.7B dense the model likely crosses EU AI Act GPAI systemic-risk thresholds that enterprise deployers should document. | Conditional 229B agent-focused model from MiniMax, Modified MIT. Strong software-engineering and tool-use benchmarks. Family has iterated fast (M2 / M2.1 / M2.5 / M2.7 across 2025-2026). Same Chinese-origin alignment and supply-chain considerations as DeepSeek, Qwen, Kimi. |
| Last reviewed | 2026-04-17 | 2026-04-16 |
| Open-weight | ||
| Licence | Apache 2.0 | Modified MIT |
| Commercial use | Unrestricted | Yes |
| Training data | Domain-level summary | Undisclosed |
| Origin | United States | China |
| Performance & pricing? | ||
| Quality index | 32/100 | 36/100 |
| Speed | — | 72 tok/s |
| Blended price | — | $0.53/M |
| Context window | — | — |
| Evidence | ||
| Sources | ||
No overlapping sources between the two entries.