Mistral Saba
Mistral Saba is a regional language model from Mistral AI, trained on curated Middle East and South Asia datasets, with Arabic and Tamil language support.
Model Overview
Capabilities, design details, and architectural traits
Mistral Saba - Specialized Regional Language Model for Middle East & South Asia
Mistral Saba is Mistral AI's first specialized regional language model. Its defining identity is not general-purpose multilingualism but custom training on meticulously curated datasets from the Middle East and South Asia a deliberate departure from general-purpose model design.
Regional Training Design
Mistral explicitly positions Saba as addressing the gap where larger general-purpose models are proficient in regional languages but lack linguistic nuances, cultural background, and in-depth regional knowledge. The model was developed in close collaboration with strategic regional customers to address specific enterprise use cases in Arabic-language and South Asian contexts.
Fine-Tuning Base
Officially documented as a strong base for training highly specific regional adaptations, particularly for domains such as energy, financial markets, and healthcare within Arabic and South Asian language and cultural contexts.
Documented Use Cases
- Conversational support: Real-time Arabic virtual assistants
- Domain-specific expertise: Fine-tuning for energy, finance, and healthcare sectors in regional language contexts
- Cultural content creation: Generating content that understands local idioms and cultural references for Middle Eastern audiences
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Mistral | Anthropic | Anthropic |
| Release Date | February 17, 2025 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | Sep 2024 | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 33K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | Free Best Input Pricing | $4 | $2 |
| Output Pricing | Free Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | textfile | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 6.5 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | - | - | - |
| Agentic Index | - | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
SciCode Benchmark
Scientific coding and mathematical modeling.
Long Context Reasoning
Logical reasoning over long context windows.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from Mistral
Other models by Mistral
Mistral Large 4 Preview
Mistral Medium 3.5
Mistral Large 3
Mistral Medium 3.1
Top AI Models
Leading alternatives by intelligence score
