Mistral Large vs Mistral Medium: Benchmark Comparison
Detailed comparison of Mistral Large and Mistral Medium covering benchmarks, pricing, context window, and compliance.
Key Specifications
| Specification | Mistral Large | Mistral Medium |
|---|---|---|
| Vendor | mistral | mistral |
| Version | large | medium |
| Release Date | 2024-02-26 | 2023-12-11 |
| Context Window | 32000 tokens | 32000 tokens |
| Input Modalities | text | text |
| Output Modalities | text | text |
| License | Apache 2.0 | Apache 2.0 |
| SOC2 | ✓ | ✓ |
| HIPAA | ✗ | ✗ |
| GDPR | ✓ | ✓ |
| ISO 27001 | ✓ | ✓ |
Benchmark Results
| Benchmark | Mistral Large | Mistral Medium | Winner |
|---|---|---|---|
| ARC | 93.3 | 94.6 | Mistral Medium |
| BBH | 76.7 | 83.4 | Mistral Medium |
| GPQA | 38.1 | 48.7 | Mistral Medium |
| GSM8K | 80.9 | 82.5 | Mistral Medium |
| HUMANEVAL | 73.4 | 78 | Mistral Medium |
| IFEVAL | 75.8 | 77.4 | Mistral Medium |
| MATH | 49.3 | 54.9 | Mistral Medium |
| MMLU | 82 | 82.9 | Mistral Medium |
| MUSR | 52.1 | 64.3 | Mistral Medium |
| WINOGRANDE | 83.3 | 82.4 | Mistral Large |
Pricing Comparison
| Tier (per Mtok) | Mistral Large | Mistral Medium |
|---|---|---|
| Input | $4 | $2.7 |
| Output | $12 | $8.1 |
| Cache Read | $0 | $0 |
| Cache Write | $0 | $0 |
Mistral Large mot Mistral Medium
Modellöversikt
Mistral Large and Mistral Medium are both notable options in the AI model market. This page compares their benchmarks, pricing, and compliance.
Nyckelspecifikationer
| Leverantör | Releasedatum | Kontextfönster | Licens |
|---|---|---|---|
| Mistral / Mistral | 2024-02-26 / 2023-12-11 | 32K / 32K | Apache 2.0 / Apache 2.0 |
Benchmarkprestanda
| Benchmark | Mistral Large | Mistral Medium | Vinnare |
|---|---|---|---|
| ARC | 93.3 | 94.6 | B |
| BBH (BIG-Bench Hard) | 76.7 | 83.4 | B |
| GPQA | 38.1 | 48.7 | B |
| GSM8K (Grade School Math 8K) | 80.9 | 82.5 | B |
| HumanEval | 73.4 | 78.0 | B |
| IFEval | 75.8 | 77.4 | B |
| MATH | 49.3 | 54.9 | B |
| MMLU (Massive Multitask Language Understanding) | 82.0 | 82.9 | B |
| MUSR | 52.1 | 64.3 | B |
| WinoGrande | 83.3 | 82.4 | A |
Prisjämförelse
| Inmatning | Utmatning | Cacheläsning | Cacheskrivning |
|---|---|---|---|
| — / — | — / — | — / — | — / — |
per miljon tokens — A / B
Styrkor & Svagheter
Mistral Large
- ✅ MMLU score 82.0, strong knowledge reasoning.
- ⚠️ 闭源专有模型，不支持自托管。
Mistral Medium
- ✅ MMLU score 82.9, strong knowledge reasoning.
- ⚠️ 闭源专有模型，不支持自托管。
Redaktörens åsikt
Mistral Large and Mistral Medium each have their strengths. Choose based on workload (code, long context, vision), referencing the tables above.
Vanliga frågor
Which model is better for coding tasks?
Refer to the HumanEval benchmark table; the model with a higher score is better suited for coding tasks.
Which model is cheaper?
Refer to the pricing comparison table above; the model with lower input/output prices is more cost-effective.
Which has a longer context window?
Refer to the key specifications table; the model with a larger context window is better for long documents.
Referenser
Editor's Take
See Editor's Take section.