MODEL COMPARISON
Compare facts.
Keep the context.
Select up to three models or families. Capabilities, licences and release dates are documented facts; family-level entries are not directly comparable to individual checkpoints.
Choose models above to update the comparison.
| Attribute | Mistral Small 3.1 | EuroLLM-9B | Viking 33B |
|---|---|---|---|
| Developer | Mistral AI | EuroLLM research collaboration | Silo AI / TurkuNLP |
| Country / region | France | Europe-wide | Finland |
| Capability category | Multimodal | Language | Language |
| Access | Open weights | Open weights | Open weights |
| Release / family | 2025-03-17 | 2025-06-04 | 2024 |
| Language coverage | Multilingual; check the model card for language-specific evaluation | 35 languages: 24 official EU languages and 11 additional languages | English, Finnish, Swedish, Norwegian, Danish, Icelandic and code |
| Context window | Up to 128k tokens (provider release documentation) | Consult the selected checkpoint model card | Consult the selected checkpoint model card |
| Licence | Apache 2.0 (Small 3.1 release) | Consult the checkpoint licence; the paper licence is separate | Check the checkpoint licence and model card |
| Profile reviewed | 16 Oct 2026 | 16 Oct 2026 | 16 Oct 2026 |
| Primary source | Open source ↗ | Open source ↗ | Open source ↗ |
An unspecified value means we have not recorded a verified value for this entry. It does not mean a model lacks the capability. Always follow the original source and evaluate the exact release for your intended task.