Mistral: Ministral 3 3B 2512

by Mistral

The smallest model in the Ministral 3 family, Ministral 3 3B is a powerful, efficient tiny language model with vision capabilities.

Avg Score

59.2%

284 answers

Avg Latency

2.9s

42 runs

Pricing

$0.10

input

$0.10

output

per 1M tokens

Context

131K

tokens

Alternatives

Models with similar or better quality but different tradeoffs

Same Quality, Cheaper

Models with similar or better performance at a lower cost per token.

Model	Cost
Meta: Llama 3.3 70B Instruct	-63%
Meta: Llama 4 Scout	-61%
Mistral: Ministral 3 8B 2512	-42%
Google: Gemini 2.0 Flash	-41%
OpenAI: gpt-oss-20b	-19%

Same Quality, Faster

Models with similar or better performance but lower latency.

Model	Latency
Google: Gemini 2.5 Flash Lite	-59%
Google: Gemini 2.0 Flash	-54%
Google: Gemini 2.5 Flash	-47%
Anthropic: Claude 3 Haiku	-40%
Mistral: Ministral 3 8B 2512	-39%

Same Cost, Better

Models at a similar price point with higher benchmark scores.

Model	Score
DeepSeek: DeepSeek V3.2	+20%
Google: Gemini 2.0 Flash	+19%
Meta: Llama 4 Maverick	+16%
Google: Gemini 2.5 Flash Lite	+14%
Meta: Llama 4 Scout	+14%

Other Models from Mistral

Compare performance with other models from the same creator

Model	Score	Latency	Cost/1M
Mistral: Mistral Large 3 2512	80.3%	5.3s	$1.00
Mistral: Ministral 3 8B 2512	70.4%	2.7s	$0.15
Mistral: Ministral 3 14B 2512	69.8%	3.9s	$0.20
Mistral: Mistral Small Creative	—	—	$0.20
Mistral: Devstral 2 2512	—	—	$0.0000
Mistral: Devstral 2 2512	—	—	$0.14
Mistral: Voxtral Small 24B 2507	—	—	$0.20
Mistral: Mistral Medium 3.1	—	—	$1.20
Mistral: Codestral 2508	—	—	$0.60
Mistral: Devstral Medium	—	—	$1.20
Mistral: Devstral Small 1.1	—	—	$0.18
Mistral: Mistral Small 3.2 24B	—	—	$0.12
Mistral: Devstral Small 2505	—	—	$0.09
Mistral: Mistral Medium 3	—	—	$1.20
Mistral: Mistral Small 3.1 24B	—	—	$0.0000
Mistral: Mistral Small 3.1 24B	—	—	$0.07
Mistral: Saba	—	—	$0.40
Mistral: Mistral Small 3	—	—	$0.07
Mistral Large 2411	—	—	$4.00
Mistral Large 2407	—	—	$4.00
Mistral: Pixtral Large 2411	—	—	$4.00
Mistral: Ministral 8B	—	—	$0.10
Mistral: Ministral 3B	—	—	$0.04
Mistral: Pixtral 12B	—	—	$0.10
Mistral: Mistral Nemo	—	—	$0.03
Mistral: Mistral 7B Instruct	—	—	$0.0000
Mistral: Mistral 7B Instruct	—	—	$0.04
Mistral: Mistral 7B Instruct v0.3	—	—	$0.20
Mistral: Mixtral 8x22B Instruct	—	—	$4.00
Mistral Large	—	—	$4.00
Mistral Tiny	—	—	$0.25
Mistral: Mistral 7B Instruct v0.2	—	—	$0.20
Mistral: Mixtral 8x7B Instruct	—	—	$0.54
Mistral: Mistral 7B Instruct v0.1	—	—	$0.15

Benchmark Performance

How this model performs across different benchmarks

Benchmark	Score	Rank
Math Bench: Subtraction (1-10)	100.0%	25 / 35
my test	100.0%	10 / 17
Customer service	100.0%	11 / 17
Testing_website	100.0%	11 / 17
cemantle solve initiator	100.0%	15 / 25
Single Character Instruction Adherence	100.0%	10 / 17
Best bathroom technique	100.0%	11 / 17
Breakfast	100.0%	4 / 17
Repetitive Onomatopoeia Response Benchmark	100.0%	10 / 17
Math Bench: Addition (1-10)	100.0%	23 / 35
Cheeky_test	100.0%	11 / 17
Customer Support Ticket Generation	100.0%	11 / 17
Smell test	100.0%	10 / 17
Enhancing FIDO UV Consistency with UXWG 2026 priority plan	95.0%	13 / 25
Motor production test	95.0%	26 / 36
Math Bench: Multiplication (1-10)	95.0%	32 / 35
Structured Customer Support Extraction	95.0%	16 / 17
SaaS Customer Support Extraction and Strategy	95.0%	14 / 17
KUKA force Torque control programming skills	90.0%	12 / 17
Instruction Adherence and Logical Constraints	90.0%	15 / 17
Define data taxonomy for product catalog suitable for PIM	80.0%	13 / 17
Math Bench: Arithmetic (1-100)	75.0%	32 / 35
Venture Capital Terms Benchmark	73.6%	23 / 25
Character Frequency Bench	71.8%	25 / 35
Math Bench: Division (1-10)	65.0%	33 / 35
moose	60.0%	13 / 17
Spatial Reasoning: Germany	58.8%	32 / 35
Product Recommendations	55.0%	31 / 36
Math Bench: Arithmetic (1-1000)	50.0%	32 / 35
Product Recommendations	43.3%	14 / 17
Docs	33.3%	8 / 17
Customer support	30.0%	17 / 17
Karlsruhe Local Knowledge Benchmark	28.9%	33 / 35
solve n-queens in clojure in one shot	20.0%	12 / 17
Omnishambles	10.0%	15 / 17
Rust performance	10.0%	13 / 16
German Memelord Bench	7.9%	26 / 35
Karl Lorey knowlege	5.6%	13 / 36
Current Global Political Leaders	0.0%	8 / 17
Money Boy Cultural Literacy Test	0.0%	32 / 35

Price vs Performance

Compare cost efficiency across all models

Current model

Other models

X-axis uses log scale for better visualization of price range

Score Over Time

Performance trends across all benchmark runs

Benchmark Activity

Number of benchmark runs over time

Quickstart

Get started with this model using OpenRouter

View on OpenRouter

import { OpenRouter } from "@openrouter/sdk";

const openrouter = new OpenRouter({
  apiKey: "<OPENROUTER_API_KEY>"
});

const completion = await openrouter.chat.completions.create({
  model: "mistralai/ministral-3b-2512",
  messages: [
    {
      role: "user",
      content: "Hello!"
    }
  ]
});

console.log(completion.choices[0].message.content);

Get your API key at openrouter.ai/keys