Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
Sibling matchup inside the LFM2.5 1.2B family.
LFM2.5-1.2B-Thinking and LFM2.5-1.2B-Instruct sit in the same LFM2.5 1.2B family. This page is less about two unrelated model lineages and more about how the siblings trade off on benchmark shape, token costs, and practical limits like context window.
LFM2.5-1.2B-Thinking has the cleaner overall profile here, landing at 33 versus 30. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
LFM2.5-1.2B-Thinking's sharpest advantage is in agentic, where it averages 34.1 against 25.7. The single biggest benchmark swing on the page is Terminal-Bench 2.0, 34 to 22. LFM2.5-1.2B-Instruct does hit back in instruction following, so the answer changes if that is the part of the workload you care about most.
LFM2.5-1.2B-Thinking is the reasoning model in the pair, while LFM2.5-1.2B-Instruct is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use.
LFM2.5-1.2B-Thinking makes more sense if agentic is the priority or you want the stronger reasoning-first profile, while LFM2.5-1.2B-Instruct is the cleaner fit if instruction following is the priority or you would rather avoid the extra latency and token burn of a reasoning model.
LFM2.5-1.2B-Thinking
34.1
LFM2.5-1.2B-Instruct
25.7
LFM2.5-1.2B-Thinking
8.2
LFM2.5-1.2B-Instruct
7.2
LFM2.5-1.2B-Thinking
32.4
LFM2.5-1.2B-Instruct
32.4
LFM2.5-1.2B-Thinking
38.4
LFM2.5-1.2B-Instruct
32.1
LFM2.5-1.2B-Thinking
27
LFM2.5-1.2B-Instruct
26
LFM2.5-1.2B-Thinking
72
LFM2.5-1.2B-Instruct
80
LFM2.5-1.2B-Thinking
60.7
LFM2.5-1.2B-Instruct
60.7
LFM2.5-1.2B-Thinking
42.3
LFM2.5-1.2B-Instruct
37
LFM2.5-1.2B-Thinking and LFM2.5-1.2B-Instruct are sibling variants in the LFM2.5 1.2B family, so the right pick depends on whether you value the better benchmark line, cheaper tokens, or the larger context window. LFM2.5-1.2B-Thinking is ahead overall 33 to 30.
LFM2.5-1.2B-Thinking has the edge for knowledge tasks in this comparison, averaging 27 versus 26. Inside this category, MMLU is the benchmark that creates the most daylight between them.
LFM2.5-1.2B-Thinking has the edge for coding in this comparison, averaging 8.2 versus 7.2. Inside this category, HumanEval is the benchmark that creates the most daylight between them.
LFM2.5-1.2B-Thinking has the edge for math in this comparison, averaging 42.3 versus 37. Inside this category, MATH-500 is the benchmark that creates the most daylight between them.
LFM2.5-1.2B-Thinking has the edge for reasoning in this comparison, averaging 38.4 versus 32.1. Inside this category, MuSR is the benchmark that creates the most daylight between them.
LFM2.5-1.2B-Thinking has the edge for agentic tasks in this comparison, averaging 34.1 versus 25.7. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.
LFM2.5-1.2B-Thinking and LFM2.5-1.2B-Instruct are effectively tied for multimodal and grounded tasks here, both landing at 32.4 on average.
LFM2.5-1.2B-Instruct has the edge for instruction following in this comparison, averaging 80 versus 72. Inside this category, IFEval is the benchmark that creates the most daylight between them.
LFM2.5-1.2B-Thinking and LFM2.5-1.2B-Instruct are effectively tied for multilingual tasks here, both landing at 60.7 on average.
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.