GLM-4.5 vs Ministral 3 3B

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

GLM-4.5 is clearly ahead on the aggregate, 36 to 27. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

GLM-4.5's sharpest advantage is in reasoning, where it averages 43.8 against 30.1. The single biggest benchmark swing on the page is MRCRv2, 52 to 35. Ministral 3 3B does hit back in multilingual, so the answer changes if that is the part of the workload you care about most.

Quick Verdict

Pick GLM-4.5 if you want the stronger benchmark profile. Ministral 3 3B only becomes the better choice if multilingual is the priority.

Agentic

GLM-4.5

GLM-4.5

31.3

Ministral 3 3B

22.9

28
Terminal-Bench 2.0
19
37
BrowseComp
33
31
OSWorld-Verified
20

Coding

GLM-4.5

GLM-4.5

14.4

Ministral 3 3B

6.2

29
HumanEval
15
18
SWE-bench Verified
8
13
LiveCodeBench
7
15
SWE-bench Pro
5

Multimodal & Grounded

GLM-4.5

GLM-4.5

41

Ministral 3 3B

30.4

36
MMMU-Pro
25
47
OfficeQA Pro
37

Reasoning

GLM-4.5

GLM-4.5

43.8

Ministral 3 3B

30.1

35
SimpleQA
22
33
MuSR
20
61
BBH
57
48
LongBench v2
32
52
MRCRv2
35

Knowledge

GLM-4.5

GLM-4.5

32

Ministral 3 3B

24.5

37
MMLU
24
36
GPQA
23
34
SuperGPQA
21
32
OpenBookQA
19
51
MMLU-Pro
48
3
HLE
1
40
FrontierScience
28

Instruction Following

GLM-4.5

GLM-4.5

68

Ministral 3 3B

67

68
IFEval
67

Multilingual

Ministral 3 3B

GLM-4.5

58.1

Ministral 3 3B

59.7

60
MGSM
61
57
MMLU-ProX
59

Mathematics

GLM-4.5

GLM-4.5

45.5

Ministral 3 3B

36

37
AIME 2023
23
39
AIME 2024
25
38
AIME 2025
24
33
HMMT Feb 2023
19
35
HMMT Feb 2024
21
34
HMMT Feb 2025
20
36
BRUMO 2025
22
57
MATH-500
53

Frequently Asked Questions

Which is better, GLM-4.5 or Ministral 3 3B?

GLM-4.5 is ahead overall, 36 to 27. The biggest single separator in this matchup is MRCRv2, where the scores are 52 and 35.

Which is better for knowledge tasks, GLM-4.5 or Ministral 3 3B?

GLM-4.5 has the edge for knowledge tasks in this comparison, averaging 32 versus 24.5. Inside this category, MMLU is the benchmark that creates the most daylight between them.

Which is better for coding, GLM-4.5 or Ministral 3 3B?

GLM-4.5 has the edge for coding in this comparison, averaging 14.4 versus 6.2. Inside this category, HumanEval is the benchmark that creates the most daylight between them.

Which is better for math, GLM-4.5 or Ministral 3 3B?

GLM-4.5 has the edge for math in this comparison, averaging 45.5 versus 36. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.

Which is better for reasoning, GLM-4.5 or Ministral 3 3B?

GLM-4.5 has the edge for reasoning in this comparison, averaging 43.8 versus 30.1. Inside this category, MRCRv2 is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, GLM-4.5 or Ministral 3 3B?

GLM-4.5 has the edge for agentic tasks in this comparison, averaging 31.3 versus 22.9. Inside this category, OSWorld-Verified is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, GLM-4.5 or Ministral 3 3B?

GLM-4.5 has the edge for multimodal and grounded tasks in this comparison, averaging 41 versus 30.4. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, GLM-4.5 or Ministral 3 3B?

GLM-4.5 has the edge for instruction following in this comparison, averaging 68 versus 67. Inside this category, IFEval is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, GLM-4.5 or Ministral 3 3B?

Ministral 3 3B has the edge for multilingual tasks in this comparison, averaging 59.7 versus 58.1. Inside this category, MMLU-ProX 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.