BenchMark'd
No ratings yet
Be the first to review this model
Mistral's mid-size model from the Ministral 3 family with vision capabilities. Scores 85% on AIME 2025 and offers excellent performance per cost. Apache 2.0 licensed with instruct, reasoning, and base variants.
Released
December 2, 2025
Parameters
14B
Context
256K
Pricing
Paid
| Benchmark | Category | Score | Performance |
|---|---|---|---|
MMLU | knowledge | 78.5% | 79 |
HumanEval | coding | 72.0% | 72 |
MATH | reasoning | 68.3% | 68 |
Chatbot Arena ELO | overall | 1215 | 59 |
Last updated: March 15, 2026
Benchmark scores may vary based on evaluation methodology and conditions.