Ministral 3 14B

Ministral 3 14B

Mistral · Released Dec 2, 2025 Professional
Intelligence #14 / 590
82.0 Our Score
Speed #171 / 279
68.5 tokens / sec
Input #277 / 592
$0.200 per 1M tokens
Output #183 / 592
$0.200 per 1M tokens
Context
— Not reported

Analysis Summary

Ministral 3 14B is the largest model in Mistral's Ministral 3 series, priced at $0.20 per million tokens. Its intelligence index of 11.1 and coding index of 14.4 place it above the 8B variant but well below Mistral Large 3. MMLU Pro of 69.3% and GPQA of 57.2% reflect competent general reasoning, and livecodebench at 35.1% suggests usable but not strong coding performance.

For businesses, this model is a reasonable middle ground for moderate-complexity tasks: content drafting, basic code generation, structured data extraction, and FAQ automation. Its agentic index of 15.9 and low terminal bench score mean it is not suited for complex agent workflows, and its instruction-following scores are modest.

At $0.20/1M tokens it is very affordable, and for teams that need more capability than the 8B but cannot justify the step up to Mistral Large 3, it offers a practical compromise. However, the capability gap between this and Mistral Large 3 is meaningful enough that teams with any serious reasoning or coding requirements should consider the larger model.

Assessed June 30, 2026

Editorial notes

Ministral 3 14B from Mistral sits between the 8B and Large 3 in capability, with improved coding and reasoning at a low $0.20/1M price, suited to moderate automation tasks.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence2.4Technical2.1Value7Content3.1
Intelligence 2.4/10
Technical 2.1/10
Content 3.1/10
Value 7/10

How Ministral 3 14B compares

Ministral 3 14B ranks #215 of 385 AI models we track for overall intelligence, #106 of 139 for coding, #214 of 293 for agentic tasks. At $0.20 per million input tokens it is cheaper than 53% of comparable models.

Performance Indices

Source: Artificial Analysis

11.1 Intelligence Index
14.4 Coding Index
15.9 Agentic Index
30 Math Index

Benchmark Scores

Intelligence

GPQA Diamond 57.2% Graduate-level scientific reasoning
HLE 4.6% Humanity's Last Exam
MMLU Pro 69.3% Multi-task language understanding
AIME 2025 30% Competition mathematics (2025)
SciCode 23.6% Scientific computing

Technical

LiveCodeBench 35.1% Live coding evaluation
TerminalBench Hard 4.5% Agentic terminal tasks
τ²-Bench 27.2% Conversational agent benchmark

Content

IFBench 32% Instruction following
LCR 22% Long-context reasoning

Benchmark data from Artificial Analysis and Hugging Face

How does Ministral 3 14B stack up?

Compare side-by-side with other professional models.

Compare Models

Model Information

ProviderMistral
Release Date December 2, 2025
Status Active

Pricing

Token Type Cost per 1M tokens Cost per 1K tokens
Input $0.20 $0.000200
Output $0.20 $0.000200

Leaderboard Categories

Frequently asked questions about Ministral 3 14B

How much does Ministral 3 14B cost?

Ministral 3 14B costs $0.20 per million input tokens and $0.20 per million output tokens.

Is Ministral 3 14B good for coding?

On our coding benchmark index, Ministral 3 14B ranks #106 of 139 models, placing it in the broader range of the field for code generation and debugging.

Who created Ministral 3 14B?

Ministral 3 14B is developed by Mistral and was released on December 2, 2025.