Side-by-side comparison of Mistral Medium 3.5 (Mistral AI · France) and QwQ-32B (Alibaba · China) for self-hosted deployment of the open-weight model. Mistral Medium 3.5 is rated EU-ready; QwQ-32B is conditional. They part ways on licence: Mistral Medium 3.5 is "Modified MIT", QwQ-32B is "Apache 2.0".
| Field | ||
|---|---|---|
| Summary | ||
| Verdict | 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. | Conditional 32B dense reasoning model under Apache 2.0. Sweet spot for self-hostable reasoning: 4090-class GPU at 4-bit, single H100 at bf16. Chinese-origin caveats unchanged. |
| Last reviewed | 2026-05-03 | 2026-04-15 |
| Open-weight | ||
| Licence | Modified MIT | Apache 2.0 |
| Commercial use | Permitted (revenue cap) | Yes |
| Training data | Undisclosed | Undisclosed |
| Origin | EU (France) | China |
| Performance & pricing? | ||
| Quality index | — | 20/100 |
| Speed | — | 33 tok/s |
| Blended price | — | $0.74/M |
| Context window | — | — |
| Evidence | ||
| Sources | ||
No overlapping sources between the two entries.