Ministral 3 8B

+ contribute

Ministral 3 8B is an open-weights language model from Mistral, released December 2025. Its published size is 8.92B parameters and a 262k-token context window. Licence: apache-2.0.

52.7±0.9#306 of 719 LLMsfrom 21 benchmark results across 8 sources

$ how it compares

The LLMs ranked either side of Ministral 3 8B, out of 828 with a score, then more from Mistral.

#	model	score	where	RAM	released
365	Phi-4 14B Microsoft	53±1	api $0.087/M +local	~10GB	2024-12
366	Gemma 4 E4B Google	53±1	local	~6.9GB	2026-04
367	Ministral 3 8B Mistral	53±1	api $0.15/M +local	~6.5GB	2025-12
368	Qwen3-VL 4B Instruct Alibaba	53±1	local	~4.1GB	2025-10
369	Claude 3 Opus Anthropic	53±0	comparison only		2024-02
	more from Mistral
79	Mistral Large 4 (public preview) Mistral	71±2	api $2.06/M		2026-10
174	Mistral Medium 3.5 (128B) Mistral	65±1	api $3.00/M +local	~42GB	2026-04
180	Mistral Medium 3.1 (2508) Mistral	64±2	comparison only		2025-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #210 of 300 * · 4.1% of its score · 1 test

Beats o1 (2024-12-17), o3-mini, DeepSeek R1 (Jan '25) here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Granite 4.1 3B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
ContractBench success rate (33 observation-contract tool tasks)	19.2% ✎	▆··▆▄▂▄▂▆▆█▄ beats 4/23	very hard · avg 14%	▇▇ 4.1%	sharp

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #285 of 393 * · 3.2% of its score · 1 test

Beats Ministral 3 14B Reasoning 2512, Gemma 4 E4B here, though they rank higher overall. Trails Flan-T5 XXL, LFM2 700M, EXAONE 3.5 2.4B Instruct here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena-Hard (version/judge unspecified) lab	50.9% ✎	▃▃▂▅▃▆▅█▅▅▂▃ beats 9/30	medium · avg 48%	▇ 3.2%	at its level · small run

math▁▂▃▅▆▅▄█▄▁▁▁ #211 of 398 · 24% of its score · 4 tests

Beats DeepSeek V3 (Dec '24), GPT-4o (2024-08-06), Command A (03-2025) here, though they rank higher overall. Trails Olmo 3 7B Instruct, DeepSeek R1 Distill Llama 8B, Phi-4-mini-reasoning 3.8B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
MATH (subset unspecified) +2	86.2% ✎	█▄··▄█▄····▄ beats 6/7	easy · avg 81%	▇▇▇▇▇ 11%	several runs
GSM8K	89.5% ✎	█····█··▄·▄█ beats 6/7	very easy · avg 86%	▇▇▇ 6.4%	sharp
PolyMath (Marco evaluation)	26.4% ✎	▄··▄██··▄··▄ beats 2/7	hard · avg 30%	▇ 3.4%	at its level
MGSM (Marco evaluation)	68.7% ✎	▄·▄·▄▄▄·▄··█ beats 4/7	easy · avg 70%	▇ 2.8%	at its level

other tests13 not in a task · 68% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA Diamond	44.8% ✎	▄···█·▄·█·▄▄ beats 6/7	hard · avg 38%	▇▇▇▇▇ 13%	sharp
Medmarks-V mean win rate (19 verifiable medical benchmarks, 71 configurations)	0.4442 ✎	▁▄▃▃▇▄█▃▅▁▁▁ beats 9/30	avg 0.423	▇▇▇▇ 11%	sharp
MMLU	79.8% ✎	▄···▄▄·▄█··█ beats 6/7	easy · avg 75%	▇▇▇ 7.7%	at its level
MMLU-Pro	63.9% ✎	▄····█·▄█·▄▄ beats 6/7	medium · avg 56%	▇▇▇ 6.3%	at its level
EuroEval Life in the UK (English knowledge MC), validation split, 5-shot, accuracy	89.7% ✎	▂·▂▂·▂·▃▃·█▃ beats 6/14	easy · avg 80%	▇▇ 5.6%	typical
Medmarks-V weighted mean win rate (61 models, 71 configurations)	44.4% ✎	▂▃▅▂▆▃█▅█▅▃▆ beats 9/34	medium · avg 42%	▇▇ 5.3%	blunt · at its level
MMLU-ProX-Lite (Marco)	43.3% ✎	█··▄··█··█·▄ beats 3/7	medium · avg 43%	▇▇ 5.2%	at its level
GlobalMMLU (Marco evaluation)	55.4% ✎	▃··▃▃··▅···█ beats 3/7	medium · avg 57%	▇ 3.7%	at its level

5 more tests

test	result	vs the others in its run	difficulty	counts	why
MMMLU (Marco evaluation)	56.4% ✎	▄···▄·█·▄·▄█ beats 2/7	medium · avg 60%	▇ 3.2%	at its level
EuroEval HellaSwag (English), validation split, 5-shot generative MC, accuracy	73.1% ✎	▂··▂▂·▂▆▆█·▄ beats 6/15	medium · avg 63%	▇ 3.1%	at its level
AGIEval (Marco suite unspecified)	52.4% ✎	▄··▄█▄··▄··█ beats 7/7	medium · avg 47%	▇ 2.5%	blunt · at its level
SurgiQ surgical MCQ (zero-shot, log-probability scoring)	54.4% ✎	▂···▂▂·▃▂▃██ beats 6/17	medium · avg 53%	▇ 1.7%	at its level
Vectara hallucination leaderboard (HHEM-2.3, factual consistency rate)	78.3% ✎	▃▂▁▂▅█▂▃▄▃▁▁ beats 3/98	very easy · avg 90%	▇ 0.0%	blunt

$ where to run it

Through an API

api · $0.15/M · $0.15 in / $0.15 out per 1M tokens · 262k context · docs.mistral.ai ↗ · price sourceUSD per 1M text tokens. Mistral standard synchronous API; model ministral-8b-2512. Uncached input; output includes generated reasoning where billed. Batch, cache storage, tools and other modalities excluded.

llama.cpp · recommended

version

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run Ministral 3 8B · needs ~5.2 GB · from huggingface.co/mistralai
chat in the terminal
llama cli -hf mistralai/Ministral-3-8B-Instruct-2512-GGUF:Q4_K_M
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf mistralai/Ministral-3-8B-Instruct-2512-GGUF:Q4_K_M

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run Ministral 3 8B · needs ~5.2 GB · from huggingface.co/mistralai
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/mistralai/Ministral-3-8B-Instruct-2512-GGUF@Q4_K_M -y
Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

Install Ollama
download it (macOS / Windows) ↗
Linux
curl -fsSL https://ollama.com/install.sh | sh
Run Ministral 3 8B · needs ~5.85 GB · from ollama.com
chat in the terminal
ollama run ministral-3:8b

MLX (Mac)

version

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run Ministral 3 8B · needs ~4.8 GB, Mac only · from huggingface.co/alexgusevski
chat in the terminal
mlx_lm.chat --model alexgusevski/Ministral-3-8B-Instruct-2512-q4-mlx
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model alexgusevski/Ministral-3-8B-Instruct-2512-q4-mlx

Weights and documentation: huggingface.co/mistralai ↗

$ how the score was worked out · Marco-MoE: Open Multilingual… 58%, Medmarks: A Comprehensive Ope… 16%, EuroEval leaderboards (result… 9% and 5 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Marco-MoE: Open Multilingual Mixture-of-Expert Language Models with Efficient Upcycling	11	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 58%	CC BY 4.0
Medmarks: A Comprehensive Open-Source LLM Benchmark Suite for Medical Tasks — authors' own runs of 61 models	2	▇▇▇▇▇ 16%	CC BY 4.0
EuroEval leaderboards (results collection), English generative leaderboard	2	▇▇▇ 9%	MIT
mistralai/Ministral-3-14B-Instruct-2512 model card · lab-reported	2	▇▇ 7%	Apache-2.0
ContractBench: Can LLM Agents Preserve Observation Contracts? (arXiv:2605.17281v2)	1	▇ 4%	CC BY 4.0
mistralai/Ministral-3-8B-Instruct-2512 model card · lab-reported	1	▇ 4%	Apache-2.0
SurgiQ: A Large-Scale Multi-Domain Benchmark for Evaluating Surgical Understanding in Large Language Models (arXiv:2606.08071v1)	1	▇ 2%	CC BY 4.0
Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository)	1	▇ 0%	Apache-2.0
prior · pull toward the middle		▇ 0%
weighted average		= 52.7

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json