LFM2-24B-A2B vs GLM-4.5-Air

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

LFM2-24B-A2B has the cleaner overall profile here, landing at 38 versus 35. It is a real lead, but still close enough that category-level strengths matter more than the headline number.

LFM2-24B-A2B's sharpest advantage is in mathematics, where it averages 50.4 against 44.4. The single biggest benchmark swing on the page is HumanEval, 42 to 27.

GLM-4.5-Air gives you the larger context window at 128K, compared with 32K for LFM2-24B-A2B.

Quick Verdict

Pick LFM2-24B-A2B if you want the stronger benchmark profile. GLM-4.5-Air only becomes the better choice if you need the larger 128K context window.

Agentic

LFM2-24B-A2B

LFM2-24B-A2B

33.4

GLM-4.5-Air

30.3

30
Terminal-Bench 2.0
28
38
BrowseComp
37
34
OSWorld-Verified
28

Coding

LFM2-24B-A2B

LFM2-24B-A2B

18

GLM-4.5-Air

14.6

42
HumanEval
27
18
SWE-bench Verified
15
17
LiveCodeBench
15
19
SWE-bench Pro
14

Multimodal & Grounded

LFM2-24B-A2B

LFM2-24B-A2B

41.7

GLM-4.5-Air

39.6

39
MMMU-Pro
36
45
OfficeQA Pro
44

Reasoning

LFM2-24B-A2B

LFM2-24B-A2B

46.6

GLM-4.5-Air

42.6

44
SimpleQA
33
42
MuSR
31
63
BBH
63
48
LongBench v2
47
45
MRCRv2
51

Knowledge

LFM2-24B-A2B

LFM2-24B-A2B

35.6

GLM-4.5-Air

31.1

46
MMLU
35
45
GPQA
34
43
SuperGPQA
32
41
OpenBookQA
30
51
MMLU-Pro
51
4
HLE
4
43
FrontierScience
37

Instruction Following

Tie

LFM2-24B-A2B

68

GLM-4.5-Air

68

68
IFEval
68

Multilingual

LFM2-24B-A2B

LFM2-24B-A2B

61.4

GLM-4.5-Air

59.1

64
MGSM
63
60
MMLU-ProX
57

Mathematics

LFM2-24B-A2B

LFM2-24B-A2B

50.4

GLM-4.5-Air

44.4

46
AIME 2023
35
48
AIME 2024
37
47
AIME 2025
36
42
HMMT Feb 2023
31
44
HMMT Feb 2024
33
43
HMMT Feb 2025
32
45
BRUMO 2025
34
57
MATH-500
57

Frequently Asked Questions

Which is better, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B is ahead overall, 38 to 35. The biggest single separator in this matchup is HumanEval, where the scores are 42 and 27.

Which is better for knowledge tasks, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B has the edge for knowledge tasks in this comparison, averaging 35.6 versus 31.1. Inside this category, MMLU is the benchmark that creates the most daylight between them.

Which is better for coding, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B has the edge for coding in this comparison, averaging 18 versus 14.6. Inside this category, HumanEval is the benchmark that creates the most daylight between them.

Which is better for math, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B has the edge for math in this comparison, averaging 50.4 versus 44.4. Inside this category, AIME 2023 is the benchmark that creates the most daylight between them.

Which is better for reasoning, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B has the edge for reasoning in this comparison, averaging 46.6 versus 42.6. Inside this category, SimpleQA is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B has the edge for agentic tasks in this comparison, averaging 33.4 versus 30.3. Inside this category, OSWorld-Verified is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B has the edge for multimodal and grounded tasks in this comparison, averaging 41.7 versus 39.6. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Which is better for instruction following, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B and GLM-4.5-Air are effectively tied for instruction following here, both landing at 68 on average.

Which is better for multilingual tasks, LFM2-24B-A2B or GLM-4.5-Air?

LFM2-24B-A2B has the edge for multilingual tasks in this comparison, averaging 61.4 versus 59.1. 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.