LFM2.5-24B-A2B measured on a Mac Studio (M5 Max, 36GB)

Results

Measure	Result	Among measured models	Note
Reply time	about 7.1 s	25 of 36	Until one simple common-sense question is answered (median of 50)
Knowledge (high school to university)	81%	13 of 36	200 questions, ±6
Math (word problems)	96%	10 of 36	100 questions, ±6
Memory used	14.9GB	21 of 36
EN→JA translation	28 pts	26 of 32	out of 100 (chrF), about 21.4 s a sentence
JA→EN translation	54 pts	20 of 32	out of 100 (chrF), about 6.4 s a sentence

"Among measured models" is the place among the 36 models measured on this site (lower is better for time and memory).

Try it on your Mac

Install the Mac version of Ollama from ollama.com and start it.
Type this in Terminal. The first time, the model downloads; then you can chat with it.ollama run hf.co/LiquidAI/LFM2.5-24B-A2B-GGUF:Q4_K_M

It used 14.9GB here. To use it comfortably alongside other apps, a Mac with 32GB of memory or more is a good guide (the model then takes about half the memory).

About the model

Maker: Liquid AI
Size: 24B parameters
Quantization: Q4_K_M
Thinks before answering: Yes
Longest input at once: 131,072 tokens
Released (Hugging Face): Oct 12, 2026
Generation speed: 151.1 tok/s
Start-up (loading): about 5.4 s
Measured on: Oct 14, 2026
Ollama: 0.34.3
Hugging Face: LiquidAI/LFM2.5-24B-A2B
Series: LFM

What the parts of a name mean (size, quantization and so on): how to read a model name.

Models of similar memory

LFM2-24B-A2B (14.8GB of memory, knowledge 53%, reply about 0.1 s)
Devstral-Small-2-24B-Instruct-2512 (16.3GB of memory, knowledge 67%, reply about 0.2 s)
Magistral-Small-2506 (16.3GB of memory, knowledge 80%, reply about 24.6 s)
Mistral-Small-3.1-24B-Instruct-2503 (16.3GB of memory, knowledge 72%, reply about 0.2 s)
Mistral-Small-3.2-24B-Instruct-2506 (16.3GB of memory, knowledge 75%, reply about 0.2 s)

Compare the main models on the text page. The series page lists every model measured.