Saba
Provider：Mistral AI
Mistral Saba is a 24B-parameter language model specifically designed for the Middle East and South Asia, delivering accurate and contextually relevant responses while maintaining efficient performance. Trained on curated regional...
Key specs
- Context：32.8K
- Max output：26.2K
- Tokenizer：Mistral
- Released：2025-02-17
Token pricing
- Input price：$0.200 / 1M tokens
- Output price：$0.600 / 1M tokens
- Cache read：$0.020 / 1M tokens
- Blended price：$0.300 / 1M tokens
Modalities
Text, File
Use Cases
- Chat & Support：FAQ auto-response, multi-turn dialogue, emotion detection — e-commerce, finance, gov.（≈ $0.0030/turn）
When to pick this model
Signals derived from public pricing and spec fields — not a benchmark. Validate with official docs and your own eval.
- Low input + output price: solid fit for high-frequency calls and batch processing.
- File / PDF input support: fit for document parsing, contract review, RAG over private data.
FAQ
- What is the token pricing of Saba? Input price $0.200 / 1M tokens, output price $0.600 / 1M tokens, blended about $0.300 / 1M tokens. Refer to Mistral AI’s official page for the exact rate.
- What context window does Saba support? Context window is 32.8K, max output about 26.2K, suitable for long documents, multi-turn chat and complex reasoning.
- What scenarios is Saba best for? Based on its capability and pricing, Saba fits best: Chat & Support. See the Use Cases section for details and cost estimates.
- What input/output modalities does Saba support? Supports Text, File modality.
- Is Saba a free model? Saba is billed per token, not a free model.
- Which provider offers Saba? Saba is offered by Mistral AI.
More models from Mistral AI
Full info：Saba