Saba
by Mistral · mistralai/mistral-saba · #84 cheapest of 325 paid models
Prices updated Jul 22, 2026, 1:33 AM UTC · refreshed hourly
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...
Compare vs:Claude Sonnet 5Claude Opus 4.6Claude Haiku 4.5GPT-5.6 Sol
Pricing
Input
$0.20
per 1M tokens
Output
$0.60
per 1M tokens
Blended (3:1)
$0.30
3 input : 1 output
Cache read
$0.02
per 1M cached tokens
Cache write
—
per 1M tokens
Cache & batch economics
Effective prices for Saba with prompt caching.
| Cache hit rate | Effective blended $/1M | Example request* | vs no cache |
|---|---|---|---|
| 0% | $0.30 | $0.021 | — |
| 50% | $0.232 | $0.012 | −43% |
| 90% | $0.178 | $0.0048 | −77% |
*Example: 100K sticky context + 2K new input + 1K output. Batch discount is an approximate 50% off list rates (available for OpenAI / Anthropic / Google — not flagged for this creator). Confirm on the vendor’s pricing page.
What a request costs
| Workload | Input tokens | Output tokens | Cost / request | Cost / 1K requests |
|---|---|---|---|---|
| Short chat message | 500 | 300 | $0.00028 | $0.28 |
| Document summary | 8,000 | 1,000 | $0.0022 | $2.20 |
| Codebase question (RAG) | 30,000 | 2,000 | $0.0072 | $7.20 |
| Long-context analysis | 150,000 | 5,000 | $0.033 | $33.00 |
Performance
Intelligence Index
6.4
AA composite quality
Coding Index
—
Math Index
—
Agentic Index
—
Output speed
0 tok/s
median
Time to first token
0s
median
Time to first answer
0s
after reasoning tokens
Artificial Analysis benchmarks
| Benchmark | Score |
|---|---|
| MMLU-Pro | 61.1% |
| GPQA Diamond | 42.4% |
| Humanity's Last Exam | 4.1% |
| SciCode | 24.1% |
| MATH-500 | 67.7% |
| AIME | 13.0% |
Specs
Context window
33K
tokens
Max output
—
tokens
Input modalities
text, file
Tokenizer
Mistral