Qwen2.5-72B vs LFM2-24B-A2B

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

Qwen2.5-72B is clearly ahead on the aggregate, 64 to 38. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Qwen2.5-72B's sharpest advantage is in mathematics, where it averages 83.5 against 50.4. The single biggest benchmark swing on the page is AIME 2023, 84 to 46.

Qwen2.5-72B gives you the larger context window at 128K, compared with 32K for LFM2-24B-A2B.

Quick Verdict

Pick Qwen2.5-72B if you want the stronger benchmark profile. LFM2-24B-A2B only becomes the better choice if its workflow or ecosystem matters more than the raw scoreboard.

Agentic

Qwen2.5-72B

Qwen2.5-72B

57.7

LFM2-24B-A2B

33.4

56
Terminal-Bench 2.0
30
64
BrowseComp
38
55
OSWorld-Verified
34

Coding

Qwen2.5-72B

Qwen2.5-72B

43.8

LFM2-24B-A2B

18

75
HumanEval
42
46
SWE-bench Verified
18
40
LiveCodeBench
17
47
SWE-bench Pro
19

Multimodal & Grounded

Qwen2.5-72B

Qwen2.5-72B

66.7

LFM2-24B-A2B

41.7

64
MMMU-Pro
39
70
OfficeQA Pro
45

Reasoning

Qwen2.5-72B

Qwen2.5-72B

75.8

LFM2-24B-A2B

46.6

80
SimpleQA
44
78
MuSR
42
81
BBH
63
72
LongBench v2
48
71
MRCRv2
45

Knowledge

Qwen2.5-72B

Qwen2.5-72B

59.8

LFM2-24B-A2B

35.6

83
MMLU
46
82
GPQA
45
80
SuperGPQA
43
78
OpenBookQA
41
75
MMLU-Pro
51
11
HLE
4
70
FrontierScience
43

Instruction Following

Qwen2.5-72B

Qwen2.5-72B

85

LFM2-24B-A2B

68

85
IFEval
68

Multilingual

Qwen2.5-72B

Qwen2.5-72B

80.8

LFM2-24B-A2B

61.4

84
MGSM
64
79
MMLU-ProX
60

Mathematics

Qwen2.5-72B

Qwen2.5-72B

83.5

LFM2-24B-A2B

50.4

84
AIME 2023
46
86
AIME 2024
48
85
AIME 2025
47
80
HMMT Feb 2023
42
82
HMMT Feb 2024
44
81
HMMT Feb 2025
43
83
BRUMO 2025
45
84
MATH-500
57

Frequently Asked Questions

Which is better, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B is ahead overall, 64 to 38. The biggest single separator in this matchup is AIME 2023, where the scores are 84 and 46.

Which is better for knowledge tasks, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for knowledge tasks in this comparison, averaging 59.8 versus 35.6. Inside this category, MMLU is the benchmark that creates the most daylight between them.

Which is better for coding, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for coding in this comparison, averaging 43.8 versus 18. Inside this category, HumanEval is the benchmark that creates the most daylight between them.

Which is better for math, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for math in this comparison, averaging 83.5 versus 50.4. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.

Which is better for reasoning, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for reasoning in this comparison, averaging 75.8 versus 46.6. Inside this category, SimpleQA is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for agentic tasks in this comparison, averaging 57.7 versus 33.4. 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, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for multimodal and grounded tasks in this comparison, averaging 66.7 versus 41.7. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for instruction following in this comparison, averaging 85 versus 68. Inside this category, IFEval is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, Qwen2.5-72B or LFM2-24B-A2B?

Qwen2.5-72B has the edge for multilingual tasks in this comparison, averaging 80.8 versus 61.4. 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.