Benchmark Almanac

Ministral 3 14B

Ministral 3 14B.

Mistral AI Released December 2nd, 2025 Family Ministral 3 The lab's model card Compare with others
22
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
Mistral AI Dec 2025

Ministral 3 14B

Open weights Small

136.2 #218 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • PHYBench 26% #5 /5
  • LMCA 12.3% #129 /147
  • KORA 6.21% #42 /42

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
Wer wird Millionär? LLM Benchmark 300.0 #54 of 77 Official leaderboard
Longform Creative Writing (EQ-Bench) 31.9 #111 of 119 Official leaderboard
Vectara Hallucination Leaderboard (2025 dataset) 19.4% #98 of 104 Official leaderboard
Structured Output Benchmark 77.8% #18 of 19 Official leaderboard

Math and science

BenchmarkScoreRankSource
AIME 2024 89.8% #20 of 88 Lab model card
AIME 2025 85% #74 of 143 Lab model card
GPQA Diamond 71.2% #176 of 302 Lab model card
DTBench 56% #150 of 180 Official leaderboard
HMMT 2025 67.5% #49 of 54 Lab model card
PHYBench 26% #5 of 5 Lab model card

Coding

BenchmarkScoreRankSource
LiveCodeBench 64.6% #18 of 54 Lab model card
LiveCodeBench v6 64.6% #70 of 90 Lab model card
DesignArena Website 1,074 #122 of 151 Official leaderboard
DesignArena Code 1,055 #119 of 143 Official leaderboard
DesignArena Game Dev 1,028 #120 of 143 Official leaderboard
DesignArena 3D Design 973 #123 of 136 Official leaderboard

Agents and tools

BenchmarkScoreRankSource
LisanBench 41.7 #126 of 133 Official leaderboard

Safety and security

BenchmarkScoreRankSource
Lakera B3 (Backbone Breaker Benchmark) 55% #17 of 35 Official leaderboard
KORA 6.21% #42 of 42 Official leaderboard

Other

BenchmarkScoreRankSource
LMCA 12.3% #129 of 147 Official leaderboard
Conceptual Reasoning Index 19.3 #137 of 151 Official leaderboard
ACCoRD 32.8 #142 of 152 Official leaderboard

Its lab's numbers put it at 136.2, number 218 of 404

The Self-Reported Index fits one score to the 6 capability results Mistral AI reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 133.9 to 138.4. How we compute it.

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Its system card

Every result in the longer documents the lab published with it. 3 of them aren't in the launch post.