Model comparison
Mistral Saba vs Mistral Large
Capability radar, value bars, and side-by-side posture.
Mistral Saba
Not scored
value score
Mistral Large
82
value score
Mistral Saba ctx
—
Mistral Large ctx
128K
Capability radar
Value · context · multimodal · openness · speed posture
Value head-to-head
- Mistral Large82
Attribute tape
| Field | Mistral Saba | Mistral Large |
|---|---|---|
| Developer | Mistral AI | Mistral AI |
| Context | — | 128K |
| Modalities | Text, Code | |
| Openness | — | Proprietary Model |
| Speed | — | Fast |
| Price | — | $2/1M in · $6/1M out |
| Value score | — | 82 |
| Summary | Mistral Saba is a 24B-parameter language model from Mistral AI, fine-tuned for Arabic and South Asian languages (including Tamil and Malayalam), designed for regional enterprise text generation and conversation with a 32k context window. It is positioned as a compact, single-GPU-deployable model intended to outperform larger general-purpose models on Middle Eastern and South Asian language tasks. | Mistral flagship API model. |