Side-by-side comparison of Gemma 4 26B A4B Instruct (Google DeepMind · United States) and Mistral Medium 3.5 (Mistral AI · France) for self-hosted deployment of the open-weight model. Gemma 4 26B A4B Instruct is rated conditional; Mistral Medium 3.5 is EU-ready. They part ways on licence: Gemma 4 26B A4B Instruct is "Apache 2.0", Mistral Medium 3.5 is "Modified MIT".
| Field | ||
|---|---|---|
| Summary | ||
| Verdict | Conditional Based on published licence terms, Gemma 4 26B A4B ships under pure Apache 2.0 with no prohibited-use carve-outs — a departure from prior Gemma generations. The sparse-MoE architecture (25.2B total / 3.8B active) puts it in an ambiguous zone for EU AI Act GPAI systemic-risk classification, and US origin plus image-input support add transparency obligations that deployers should document. | EU-ready Per Mistral's published license, Mistral Medium 3.5 ships under a Modified MIT License: vanilla-MIT terms for independent developers, startups and most enterprises, with a revenue-threshold carve-out that routes large companies back to a paid commercial agreement. Paris-headquartered vendor with EU jurisdiction end to end and a 256k-token context. Training-data composition is not published in the model card — usual caveat for high-assurance public-sector deployments. |
| Last reviewed | 2026-04-17 | 2026-05-03 |
| Open-weight | ||
| Licence | Apache 2.0 | Modified MIT |
| Commercial use | Unrestricted | Permitted (revenue cap) |
| Training data | Domain-level summary | Undisclosed |
| Origin | United States | EU (France) |
| Performance & pricing? | ||
| Quality index | 27/100 | — |
| Speed | — | — |
| Blended price | — | — |
| Context window | — | — |
| Evidence | ||
| Sources | ||
No overlapping sources between the two entries.