Model Details
Mistral Medium 3.5
Quick answer: Mistral Medium 3.5 is Mistral AI's October 2025 Medium tier model scoring 77.6% SWE-Bench Verified and 4.52% HLE. Medium tier — between Small and Large.
At a Glance
Where Mistral Medium 3.5 leads
- 77.6% SWE-Bench Verified — competitive software engineering
Where it lags
- 4.52% HLE — very low frontier reasoning
- Proprietary license
Best for: Mistral Medium API SWE-Bench workloads; code generation at medium tier.
What Mistral Medium 3.5 Is
Mistral Medium 3.5 is Mistral AI's October 2025 medium-tier API model. The 77.6% SWE-Bench Verified is strong (close to Qwen3.7 Plus at 77.7%), but the 4.52% HLE is unusually low for an October 2025 model.
Specifications
| Field | Value |
|---|---|
| Organization | Mistral AI |
| License | Mistral Medium 3.5 License |
| Release date | October 2025 |
| Modality | Text only |
Pricing
Available via Mistral AI API.
Public Benchmark Scores
| Benchmark | Score | Source | Date |
|---|---|---|---|
| SWE-Bench Verified | 77.6% | Benchgen evaluation | 2025-10 |
| Humanity's Last Exam | 4.52% | Benchgen evaluation | 2025-10 |
Mistral Medium 3.5 vs Alternatives
| Model | SWE-Bench Verified | HLE | License |
|---|---|---|---|
| Mistral Medium 3.5 | 77.6% | 4.52% | Proprietary |
| Mistral Large 3 | — | — | Commercial |
| Qwen3.7 Plus | 77.7% | 34.7% | Proprietary |
For SWE-Bench at similar level (77.7%) with better HLE (34.7%): Qwen3.7 Plus.
Related Models & Benchmarks
- Compare: Mistral Large 3 · Qwen3.7 Plus
- Benchmarks: SWE-Bench Verified · Humanity's Last Exam
Frequently Asked Questions
What is Mistral Medium 3.5?
Mistral AI's October 2025 medium tier scoring 77.6% SWE-Bench Verified. Oct 2025.Specs from Mistral AI's Mistral Medium 3.5 release (October 2025) and Benchgen evaluations. Last updated 2026-07-24.