Magistral Small 1.2 vs Magistral Medium 1.2
Comparing 2 AI models · 5 benchmarks · Mistral
Composite Indices
Intelligence, Coding, Math
Standard Benchmarks
Academic and industry benchmarks
Benchmark Winners
Magistral Small 1.2
No clear wins
Magistral Medium 1.2
- GPQA
- MMLU Pro
- HLE
- LiveCodeBench
- AIME 2025
| Metric | Mi Magistral Small 1.2 | Mi Magistral Medium 1.2 |
|---|---|---|
| Pricing Per 1M tokens | ||
| Input Cost | $0.50/1M | $2.00/1M |
| Output Cost | $1.50/1M | $5.00/1M |
| Blended Cost 3:1 input/output ratio | $0.75/1M | $2.75/1M |
| Specifications | ||
| Organization Model creator | Mistral | Mistral |
| Release Date Launch date | Sep 17, 2025 | Sep 18, 2025 |
| Performance & Speed | ||
| Throughput Output speed | 194.0 tok/s | 95.0 tok/s |
| Time to First Token (TTFT) Initial response delay | 352ms | 456ms |
| Latency Time to first answer token | 10664ms | 21506ms |
| Composite Indices | ||
| Intelligence Index Overall reasoning capability | 43.0 | 52.0 |
| Coding Index Programming ability | 37.2 | 42.3 |
| Math Index Mathematical reasoning | 80.3 | 82.0 |
| Standard Benchmarks | ||
| GPQA Graduate-level reasoning | 66.3% | 73.9% |
| MMLU Pro Advanced knowledge | 76.8% | 81.5% |
| HLE Hard language evaluation | 6.1% | 9.6% |
| LiveCodeBench Real-world coding tasks | 72.3% | 75.0% |
| MATH 500 Mathematical problems | — | — |
| AIME 2025 Advanced math competition | 80.3% | 82.0% |
| AIME (Original) Math olympiad problems | — | — |
| SciCode Scientific code generation | 35.2% | 39.2% |
| LCR Code review capability | 16.3% | 51.3% |
| IFBench Instruction-following | 44.4% | 43.0% |
| TAU-bench v2 Tool use & agentic tasks | 27.8% | 52.0% |
| TerminalBench Hard CLI command generation | 4.3% | 12.8% |
Key Takeaways
Magistral Small 1.2 offers the best value at $0.50/1M, making it ideal for high-volume applications and cost-conscious projects.
Magistral Medium 1.2 leads in reasoning capabilities with a 73.9% GPQA score, excelling at complex analytical tasks and problem-solving.
Magistral Medium 1.2 achieves a 42.3 coding index, making it the top choice for software development and code generation tasks.
All models support context windows of ∞+ tokens, suitable for processing lengthy documents and maintaining extended conversations.
When to Choose Each Model
Magistral Small 1.2
- Cost-sensitive applications
- High-volume processing
Magistral Medium 1.2
- Complex reasoning tasks
- Research & analysis
- Code generation
- Software development
Try Models for Free
Try Magistral Small 1.2 for FREE
No credit card or account required.
Try Magistral Medium 1.2 for FREE
No credit card or account required.
Cost Calculator
Magistral Small 1.2 Magistral Medium 1.2