Ministral 3 8B vs Ministral 3 3B

Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.

Ministral 3 8B is clearly ahead on the aggregate, 32 to 27. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Ministral 3 8B's sharpest advantage is in coding, where it averages 14.2 against 6.2. The single biggest benchmark swing on the page is AIME 2025, 33 to 24.

Quick Verdict

Pick Ministral 3 8B if you want the stronger benchmark profile. Ministral 3 3B only becomes the better choice if its workflow or ecosystem matters more than the raw scoreboard.

Agentic

Ministral 3 8B

Ministral 3 8B

28.9

Ministral 3 3B

22.9

26
Terminal-Bench 2.0
19
36
BrowseComp
33
27
OSWorld-Verified
20

Coding

Ministral 3 8B

Ministral 3 8B

14.2

Ministral 3 3B

6.2

23
HumanEval
15
16
SWE-bench Verified
8
15
LiveCodeBench
7
13
SWE-bench Pro
5

Multimodal & Grounded

Ministral 3 8B

Ministral 3 8B

32.4

Ministral 3 3B

30.4

27
MMMU-Pro
25
39
OfficeQA Pro
37

Reasoning

Ministral 3 8B

Ministral 3 8B

36.1

Ministral 3 3B

30.1

28
SimpleQA
22
26
MuSR
20
63
BBH
57
38
LongBench v2
32
41
MRCRv2
35

Knowledge

Ministral 3 8B

Ministral 3 8B

28

Ministral 3 3B

24.5

28
MMLU
24
27
GPQA
23
25
SuperGPQA
21
23
OpenBookQA
19
52
MMLU-Pro
48
3
HLE
1
32
FrontierScience
28

Instruction Following

Ministral 3 8B

Ministral 3 8B

69

Ministral 3 3B

67

69
IFEval
67

Multilingual

Ministral 3 8B

Ministral 3 8B

61.7

Ministral 3 3B

59.7

63
MGSM
61
61
MMLU-ProX
59

Mathematics

Ministral 3 8B

Ministral 3 8B

43.3

Ministral 3 3B

36

30
AIME 2023
23
32
AIME 2024
25
33
AIME 2025
24
26
HMMT Feb 2023
19
28
HMMT Feb 2024
21
27
HMMT Feb 2025
20
29
BRUMO 2025
22
60
MATH-500
53

Frequently Asked Questions

Which is better, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B is ahead overall, 32 to 27. The biggest single separator in this matchup is AIME 2025, where the scores are 33 and 24.

Which is better for knowledge tasks, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for knowledge tasks in this comparison, averaging 28 versus 24.5. Inside this category, MMLU is the benchmark that creates the most daylight between them.

Which is better for coding, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for coding in this comparison, averaging 14.2 versus 6.2. Inside this category, HumanEval is the benchmark that creates the most daylight between them.

Which is better for math, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for math in this comparison, averaging 43.3 versus 36. Inside this category, AIME 2025 is the benchmark that creates the most daylight between them.

Which is better for reasoning, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for reasoning in this comparison, averaging 36.1 versus 30.1. Inside this category, SimpleQA is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for agentic tasks in this comparison, averaging 28.9 versus 22.9. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for multimodal and grounded tasks in this comparison, averaging 32.4 versus 30.4. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for instruction following in this comparison, averaging 69 versus 67. Inside this category, IFEval is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, Ministral 3 8B or Ministral 3 3B?

Ministral 3 8B has the edge for multilingual tasks in this comparison, averaging 61.7 versus 59.7. Inside this category, MGSM is the benchmark that creates the most daylight between them.

Last updated: March 12, 2026

Weekly LLM Benchmark Digest

Get notified when new models drop, benchmark scores change, or the leaderboard shifts. One email per week.

Free. No spam. Unsubscribe anytime. We only store derived location metadata for consent routing.