Ministral-3-14B-Instruct-2512

<h1 align: center; style="display: flex; align-items: center; gap: 10px; margin: 0;"> Ministral 3 14B Instruct 2512 <img src="https://www.redhat.com/rhdc/managed-files/Catalog-Validated_model_0.png" alt="Model Icon" width="40" style="margin: 0; padding: 0;" /> </h1> <a href="https://www.redhat.com/en/products/ai/validated-models" target="_blank" style="margin: 0; padding: 0;"> <img src="https://www.redhat.com/rhdc/managed-files/Validated_badge-Dark.png" alt="Validated Badge" width="250" style="margin: 0; padding: 0;" /> </a>

The largest model in the Ministral 3 family, Ministral 3 14B offers frontier capabilities and performance comparable to its larger [Mistral Small 3.2 24B](https://huggingface.co/mistralai/ Ministral-3-14B-Instruct-2512) counterpart. A powerful and efficient language model with vision capabilities.

This model is the instruct post-trained version in FP8, fine-tuned for instruction tasks, making it ideal for chat and instruction based use cases.

The Ministral 3 family is designed for edge deployment, capable of running on a wide range of hardware. Ministral 3 14B can even be deployed locally, capable of fitting in 24GB of VRAM in FP8, and less if further quantized.

Learn more in our blog post here.

ModelCar Storage URI: oci://registry.redhat.io/rhai/modelcar-ministral-3-14b-instruct-2512:3.0

Validated on vLLM: 0.13.0

Validated on RHAIIS: 3.3

Validated on RHOAI: 3.3

Key Features

Ministral 3 14B consists of two main architectural components:

• 13.5B Language Model
• 0.4B Vision Encoder

The Ministral 3 14B Instruct model offers the following capabilities:

• Vision: Enables the model to analyze images and provide insights based on visual content, in addition to text.
• Multilingual: Supports dozens of languages, including English, French, Spanish, German, Italian, Portuguese, Dutch, Chinese, Japanese, Korean, Arabic.
• System Prompt: Maintains strong adherence and support for system prompts.
• Agentic: Offers best-in-class agentic capabilities with native function calling and JSON outputting.
• Edge-Optimized: Delivers best-in-class performance at a small scale, deployable anywhere.
• Apache 2.0 License: Open-source license allowing usage and modification for both commercial and non-commercial purposes.
• Large Context Window: Supports a 256k context window.

Use Cases

Private AI deployments where advanced capabilities meet practical hardware constraints:

• Private/custom chat and AI assistant deployments in constrained environments
• Advanced local agentic use cases
• Fine-tuning and specialization
• And more...

Bringing advanced AI capabilities to most environments.

Recommended Settings

We recommend deploying with the following best practices:

• System Prompt: Define a clear environment and use case, including guidance on how to effectively leverage tools in agentic systems.
• Sampling Parameters: Use a temperature below 0.1 for daily-driver and production environments ; Higher temperatures may be explored for creative use cases - developers are encouraged to experiment with alternative settings.
• Tools: Keep the set of tools well-defined and limit their number to the minimum required for the use case - Avoiding overloading the model with an excessive number of tools.
• Vision: When deploying with vision capabilities, we recommend maintaining an aspect ratio close to 1:1 (width-to-height) for images. Avoiding the use of overly thin or wide images - crop them as needed to ensure optimal performance.

Ministral 3 Family

Other formats available here.

Benchmark Results

We compare Ministral 3 to similar sized models.

Reasoning

Instruct

Base

Usage

The model can be used with the following frameworks;

• vllm: See here
• transformers: See here

vLLM

We recommend using this model with vLLM.

Installation

Make sure to install vllm >= 0.12.0:

pip install vllm --upgrade

Doing so should automatically install mistral_common >= 1.8.6.

To check:

python -c "import mistral_common; print(mistral_common.__version__)"

You can also make use of a ready-to-go docker image or on the docker hub.

Serve

Due to their size and the FP8 format of their weights Ministral-3-3B-Instruct-2512, Ministral-3-8B-Instruct-2512 and Ministral-3-14B-Instruct-2512 can run on a single 1xH200 GPU.

A simple launch command is:

vllm serve mistralai/Ministral-3-14B-Instruct-2512 \
  --tokenizer_mode mistral --config_format mistral --load_format mistral \
  --enable-auto-tool-choice --tool-call-parser mistral

Key parameter notes:

• enable-auto-tool-choice: Required when enabling tool usage.
• tool-call-parser mistral: Required when enabling tool usage.

Additional flags:

• You can set --max-model-len to preserve memory. By default it is set to 262144 which is quite large but not necessary for most scenarios.
• You can set --max-num-batched-tokens to balance throughput and latency, higher means higher throughput but higher latency.

Usage of the model

Here we assume that the model mistralai/Ministral-3-14B-Instruct-2512 is served and you can ping it to the domain localhost with the port 8000 which is the default for vLLM.

<details> <summary>Vision Reasoning</summary>

Let's see if the Ministral 3 knows when to pick a fight !

from datetime import datetime, timedelta

from openai import OpenAI
from huggingface_hub import hf_hub_download

# Modify OpenAI's API key and API base to use vLLM's API server.
openai_api_key = "EMPTY"
openai_api_base = "http://localhost:8000/v1"

TEMP = 0.15
MAX_TOK = 262144

client = OpenAI(
    api_key=openai_api_key,
    base_url=openai_api_base,
)

models = client.models.list()
model = models.data[0].id


def load_system_prompt(repo_id: str, filename: str) -> str:
    file_path = hf_hub_download(repo_id=repo_id, filename=filename)
    with open(file_path, "r") as file:
        system_prompt = file.read()
    today = datetime.today().strftime("%Y-%m-%d")
    yesterday = (datetime.today() - timedelta(days=1)).strftime("%Y-%m-%d")
    model_name = repo_id.split("/")[-1]
    return system_prompt.format(name=model_name, today=today, yesterday=yesterday)


SYSTEM_PROMPT = load_system_prompt(model, "SYSTEM_PROMPT.txt")
image_url = "https://static.wikia.nocookie.net/essentialsdocs/images/7/70/Battle.png/revision/latest?cb=20220523172438"

messages = [
    {"role": "system", "content": SYSTEM_PROMPT},
    {
        "role": "user",
        "content": [
            {
                "type": "text",
                "text": "What action do you think I should take in this situation? List all the possible actions and explain why you think they are good or bad.",
            },
            {"type": "image_url", "image_url": {"url": image_url}},
        ],
    },
]


response = client.chat.completions.create(
    model=model,
    messages=messages,
    temperature=TEMP,
    max_tokens=MAX_TOK,
)

print(response.choices[0].message.content)

</details> <details> <summary>Function Calling</summary>

Let's solve some equations thanks to our simple Python calculator tool.

import json
from openai import OpenAI
from huggingface_hub import hf_hub_download

# Modify OpenAI's API key and API base to use vLLM's API server.
openai_api_key = "EMPTY"
openai_api_base = "http://localhost:8000/v1"

TEMP = 0.15
MAX_TOK = 262144

client = OpenAI(
    api_key=openai_api_key,
    base_url=openai_api_base,
)

models = client.models.list()
model = models.data[0].id


def load_system_prompt(repo_id: str, filename: str) -> str:
    file_path = hf_hub_download(repo_id=repo_id, filename=filename)
    with open(file_path, "r") as file:
        system_prompt = file.read()
    return system_prompt


SYSTEM_PROMPT = load_system_prompt(model, "SYSTEM_PROMPT.txt")

image_url = "https://math-coaching.com/img/fiche/46/expressions-mathematiques.jpg"


def my_calculator(expression: str) -> str:
    return str(eval(expression))


tools = [
    {
        "type": "function",
        "function": {
            "name": "my_calculator",
            "description": "A calculator that can evaluate a mathematical expression.",
            "parameters": {
                "type": "object",
                "properties": {
                    "expression": {
                        "type": "string",
                        "description": "The mathematical expression to evaluate.",
                    },
                },
                "required": ["expression"],
            },
        },
    },
    {
        "type": "function",
        "function": {
            "name": "rewrite",
            "description": "Rewrite a given text for improved clarity",
            "parameters": {
                "type": "object",
                "properties": {
                    "text": {
                        "type": "string",
                        "description": "The input text to rewrite",
                    }
                },
            },
        },
    },
]

messages = [
    {"role": "system", "content": SYSTEM_PROMPT},
    {
        "role": "user",
        "content": [
            {
                "type": "text",
                "text": "Thanks to your calculator, compute the results for the equations that involve numbers displayed in the image.",
            },
            {
                "type": "image_url",
                "image_url": {
                    "url": image_url,
                },
            },
        ],
    },
]

response = client.chat.completions.create(
    model=model,
    messages=messages,
    temperature=TEMP,
    max_tokens=MAX_TOK,
    tools=tools,
    tool_choice="auto",
)

tool_calls = response.choices[0].message.tool_calls

results = []
for tool_call in tool_calls:
    function_name = tool_call.function.name
    function_args = tool_call.function.arguments
    if function_name == "my_calculator":
        result = my_calculator(**json.loads(function_args))
        results.append(result)

messages.append({"role": "assistant", "tool_calls": tool_calls})
for tool_call, result in zip(tool_calls, results):
    messages.append(
        {
            "role": "tool",
            "tool_call_id": tool_call.id,
            "name": tool_call.function.name,
            "content": result,
        }
    )


response = client.chat.completions.create(
    model=model,
    messages=messages,
    temperature=TEMP,
    max_tokens=MAX_TOK,
)

print(response.choices[0].message.content)

</details> <details> <summary>Text-Only Request</summary>

Ministral 3 can follow your instructions to the letter.

from openai import OpenAI
from huggingface_hub import hf_hub_download

# Modify OpenAI's API key and API base to use vLLM's API server.
openai_api_key = "EMPTY"
openai_api_base = "http://localhost:8000/v1"

TEMP = 0.15
MAX_TOK = 262144

client = OpenAI(
    api_key=openai_api_key,
    base_url=openai_api_base,
)

models = client.models.list()
model = models.data[0].id


def load_system_prompt(repo_id: str, filename: str) -> str:
    file_path = hf_hub_download(repo_id=repo_id, filename=filename)
    with open(file_path, "r") as file:
        system_prompt = file.read()
    return system_prompt


SYSTEM_PROMPT = load_system_prompt(model, "SYSTEM_PROMPT.txt")

messages = [
    {"role": "system", "content": SYSTEM_PROMPT},
    {
        "role": "user",
        "content": "Write me a sentence where every word starts with the next letter in the alphabet - start with 'a' and end with 'z'.",
    },
]

response = client.chat.completions.create(
    model=model,
    messages=messages,
    temperature=TEMP,
    max_tokens=MAX_TOK,
)

assistant_message = response.choices[0].message.content
print(assistant_message)

</details>

Transformers

You can also use Ministral 3 14B Instruct 2512 with Transformers !

Transformers recently added support for FP8, so make sure to install from main:

uv pip install git+https://github.com/huggingface/transformers

To make the best use of our model with Transformers make sure to have installed mistral-common >= 1.8.6 to use our tokenizer.

pip install mistral-common --upgrade

Try it out by running the following snippet.

[!Tip] On latest main as of 05/12/2025, by default a FP8 triton kernel for fast accelerated matmuls (w8a8_block_fp8_matmul_triton) will be used without any degradation in accuracy. However, if you want to run your model in BF16 see (here)

<details> <summary>Python snippet</summary>

import torch
from transformers import Mistral3ForConditionalGeneration, MistralCommonBackend

model_id = "mistralai/Ministral-3-14B-Instruct-2512"

tokenizer = MistralCommonBackend.from_pretrained(model_id)
model = Mistral3ForConditionalGeneration.from_pretrained(model_id, device_map="auto")

image_url = "https://static.wikia.nocookie.net/essentialsdocs/images/7/70/Battle.png/revision/latest?cb=20220523172438"

messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "text",
                "text": "What action do you think I should take in this situation? List all the possible actions and explain why you think they are good or bad.",
            },
            {"type": "image_url", "image_url": {"url": image_url}},
        ],
    },
]

tokenized = tokenizer.apply_chat_template(messages, return_tensors="pt", return_dict=True)

tokenized["input_ids"] = tokenized["input_ids"].to(device="cuda")
tokenized["pixel_values"] = tokenized["pixel_values"].to(dtype=torch.bfloat16, device="cuda")
image_sizes = [tokenized["pixel_values"].shape[-2:]]

output = model.generate(
    **tokenized,
    image_sizes=image_sizes,
    max_new_tokens=512,
)[0]

decoded_output = tokenizer.decode(output[len(tokenized["input_ids"][0]):])
print(decoded_output)

</details>

Transformers BF16

Transformers allows you to automatically convert the checkpoint to Bfloat16. To do so, simply load the model as follows:

from transformers import Mistral3ForConditionalGeneration, FineGrainedFP8Config

model_id = "mistralai/Ministral-3-14B-Instruct-2512"
model = Mistral3ForConditionalGeneration.from_pretrained(
    model_id,
    device_map="auto",
    quantization_config=FineGrainedFP8Config(dequantize=True)
)

Red Hat AI Evaluations

As part of the model validation effort, Red Hat conducted independent accuracy evaluations and the results are presented below. The model was evaluated with vLLM version 0.11.2 and either lm-evaluation-harness or lighteval depending on the benchmark.

<details> <summary>Evaluation commands</summary>

All evaluations were conducted using the vLLM server interface. The server is first initialized with the following command on 1 H200 GPUs:

vllm serve RedHatAI/Ministral-3-14B-Instruct-2512 \
  --max-model-len 262144 \
  --tokenizer_mode mistral \
  --config_format mistral \
  --load_format mistral \
  --limit-mm-per-prompt '{"image": 10}'

MMLU-Pro, IFEval and MMMU were evaluated using lm-evaluation-harness as follows.

lm_eval \
  --model local-chat-completions \
  --tasks mmlu_pro,ifeval,mmmu_val \
  --model_args "model=RedHatAI/Ministral-3-14B-Instruct-2512,max_length=64000,base_url=http://0.0.0.0:8000/v1/chat/completions,num_concurrent=64,max_retries=3,tokenized_requests=False,tokenizer_backend=None,timeout=1200,max_images=10" \
  --apply_chat_template \
  --fewshot_as_multiturn \
  --output_path results_lmeval_ministral \
  --gen_kwargs "do_sample=True,temperature=0.15"

AIME25, GPQA Diamond and Math 500 were evaluated using lighteval as follows.

litellm_config.yaml

model_parameters:
  provider: "hosted_vllm"
  model_name: "hosted_vllm/RedHatAI/Ministral-3-14B-Instruct-2512"
  base_url: "http://0.0.0.0:8000/v1"
  api_key: ""
  timeout: 1200
  concurrent_requests: 64
  max_model_length: 262144
  generation_parameters:
    temperature: 0.15
    max_new_tokens: 64000

lighteval endpoint litellm litellm_config.yaml \
"aime25|0,math_500|0,gpqa:diamond|0" \
--output-dir results_lighteval_ministral \
--save-details

</details> <table> <thead> <tr> <th>Benchmark</th> <th>RedHatAI/Ministral-3-14B-Instruct-2512-BF16</th> <th>RedHatAI/Ministral-3-14B-Instruct-2512</th> <th>Recovery</th> </tr> </thead> <tbody> <tr> <td>MMLU-Pro</td> <td>41.69</td> <td>45.87</td> <td>110.0%</td> </tr> <tr> <td>IFEval</td> <td>77.34</td> <td>76.86</td> <td>99.38%</td> </tr> <tr> <td>MMMU</td> <td>55.33</td> <td>55.33</td> <td>100.0%</td> </tr> <tr> <td>AIME25</td> <td>36.67</td> <td>36.67</td> <td>100.0%</td> </tr> <tr> <td>GPQA Diamond</td> <td>58.59</td> <td>58.59</td> <td>100.0%</td> </tr> <tr> <td>MATH 500</td> <td>88.6</td> <td>86.2</td> <td>97.29%</td> </tr> </tbody> </table> ## License

This model is licensed under the Apache 2.0 License.

You must not use this model in a manner that infringes, misappropriates, or otherwise violates any third party’s rights, including intellectual property rights.