@artyomita: Four tiny AI models, 1 GB each, against rivals 4 to 6 times their size. LFM3-1.7B wins 14 of 29 tests vs Qwen3.5-9B. LFM3-VL-2B wins 28 of 62 vs Gemma 4 12B. LFM3-Omni-2B wins 6 of 10 vs Qwen2.5-Omni 10.7B. LFM3-Audio-2B wins 4 of 5 vs Gemma 4 E4B. All counted from the model cards. They run on llama.cpp. Which one should I test first? Code: github.com/ArtyomITA Music: "Late Night Radio" by Kevin MacLeod (incompetech.com), CC BY 4.0 #ai #llm #localai #liquidai