Qwen3.5 397B vs Ministral 3 3B

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

Qwen3.5 397B is clearly ahead on the aggregate, 62 to 27. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Qwen3.5 397B's sharpest advantage is in reasoning, where it averages 75.9 against 30.1. The single biggest benchmark swing on the page is HumanEval, 75 to 15.

Quick Verdict

Pick Qwen3.5 397B 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

Qwen3.5 397B

Qwen3.5 397B

56.9

Ministral 3 3B

22.9

58
Terminal-Bench 2.0
19
62
BrowseComp
33
52
OSWorld-Verified
20

Coding

Qwen3.5 397B

Qwen3.5 397B

40.7

Ministral 3 3B

6.2

75
HumanEval
15
42
SWE-bench Verified
8
39
LiveCodeBench
7
42
SWE-bench Pro
5

Multimodal & Grounded

Qwen3.5 397B

Qwen3.5 397B

61.4

Ministral 3 3B

30.4

56
MMMU-Pro
25
68
OfficeQA Pro
37

Reasoning

Qwen3.5 397B

Qwen3.5 397B

75.9

Ministral 3 3B

30.1

80
SimpleQA
22
78
MuSR
20
82
BBH
57
72
LongBench v2
32
71
MRCRv2
35

Knowledge

Qwen3.5 397B

Qwen3.5 397B

59.3

Ministral 3 3B

24.5

83
MMLU
24
82
GPQA
23
80
SuperGPQA
21
78
OpenBookQA
19
73
MMLU-Pro
48
10
HLE
1
71
FrontierScience
28

Instruction Following

Qwen3.5 397B

Qwen3.5 397B

82

Ministral 3 3B

67

82
IFEval
67

Multilingual

Qwen3.5 397B

Qwen3.5 397B

78.8

Ministral 3 3B

59.7

82
MGSM
61
77
MMLU-ProX
59

Mathematics

Qwen3.5 397B

Qwen3.5 397B

81.6

Ministral 3 3B

36

83
AIME 2023
23
85
AIME 2024
25
84
AIME 2025
24
79
HMMT Feb 2023
19
81
HMMT Feb 2024
21
80
HMMT Feb 2025
20
82
BRUMO 2025
22
81
MATH-500
53

Frequently Asked Questions

Which is better, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B is ahead overall, 62 to 27. The biggest single separator in this matchup is HumanEval, where the scores are 75 and 15.

Which is better for knowledge tasks, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for knowledge tasks in this comparison, averaging 59.3 versus 24.5. Inside this category, MMLU is the benchmark that creates the most daylight between them.

Which is better for coding, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for coding in this comparison, averaging 40.7 versus 6.2. Inside this category, HumanEval is the benchmark that creates the most daylight between them.

Which is better for math, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for math in this comparison, averaging 81.6 versus 36. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.

Which is better for reasoning, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for reasoning in this comparison, averaging 75.9 versus 30.1. Inside this category, SimpleQA is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for agentic tasks in this comparison, averaging 56.9 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, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for multimodal and grounded tasks in this comparison, averaging 61.4 versus 30.4. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for instruction following in this comparison, averaging 82 versus 67. Inside this category, IFEval is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, Qwen3.5 397B or Ministral 3 3B?

Qwen3.5 397B has the edge for multilingual tasks in this comparison, averaging 78.8 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.