Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Phi-4 is clearly ahead on the aggregate, 42 to 38. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Phi-4's sharpest advantage is in coding, where it averages 55 against 18. The single biggest benchmark swing on the page is HumanEval, 82.6 to 42. LFM2-24B-A2B does hit back in reasoning, so the answer changes if that is the part of the workload you care about most.
LFM2-24B-A2B is also the more expensive model on tokens at $0.03 input / $0.12 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for Phi-4. That is roughly Infinityx on output cost alone. LFM2-24B-A2B gives you the larger context window at 32K, compared with 16K for Phi-4.
Pick Phi-4 if you want the stronger benchmark profile. LFM2-24B-A2B only becomes the better choice if reasoning is the priority or you need the larger 32K context window.
Phi-4
38.3
LFM2-24B-A2B
33.4
Phi-4
55
LFM2-24B-A2B
18
Phi-4
46.8
LFM2-24B-A2B
41.7
Phi-4
31.3
LFM2-24B-A2B
46.6
Phi-4
53.8
LFM2-24B-A2B
35.6
Comparable scores for this category are coming soon. One or both models do not have sourced results here yet.
Phi-4
67.2
LFM2-24B-A2B
61.4
Comparable scores for this category are coming soon. One or both models do not have sourced results here yet.
Phi-4 is ahead overall, 42 to 38. The biggest single separator in this matchup is HumanEval, where the scores are 82.6 and 42.
Phi-4 has the edge for knowledge tasks in this comparison, averaging 53.8 versus 35.6. Inside this category, MMLU is the benchmark that creates the most daylight between them.
Phi-4 has the edge for coding in this comparison, averaging 55 versus 18. Inside this category, HumanEval is the benchmark that creates the most daylight between them.
LFM2-24B-A2B has the edge for reasoning in this comparison, averaging 46.6 versus 31.3. Inside this category, LongBench v2 is the benchmark that creates the most daylight between them.
Phi-4 has the edge for agentic tasks in this comparison, averaging 38.3 versus 33.4. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.
Phi-4 has the edge for multimodal and grounded tasks in this comparison, averaging 46.8 versus 41.7. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.
Phi-4 has the edge for multilingual tasks in this comparison, averaging 67.2 versus 61.4. Inside this category, MGSM is the benchmark that creates the most daylight between them.
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.