What’s the point of smaller models?

<!-- SC_OFF --><div class="md"><p>Basically title. All labs seem to have agreed 24-27b is the minimum size they will go, nothing below that. I wonder when will we get the next sub

<!-- SC_OFF --><div class="md"><p>What models do you think are best for UI/design? I saw that GLM5.1 was quite capable, Opus and Sonnet as well, but are there any smaller models

<!-- SC_OFF --><div class="md"><p>Hi. What do you think is the smallest Open model, as of Spring 2026, that can match/beat OpenAI GPT 4o Mini in the broadest sense possible?</p>

<!-- SC_OFF --><div class="md"><p>GPT-OSS-120B was the first model of that family, which was followed by GLM-4.5-Air, Nemotron-3-Super, Qwen3.5-122B, Mistral-Small-4-119B.

<!-- SC_OFF --><div class="md"><p>I’m trying to understand the current open-source LLM landscape beyond surface-level hype.</p> <p>We all got used to the nerfed products of

<!-- SC_OFF --><div class="md"><p>I was messing with a 3B parameter model last night and it seemed way less prone to hallucinating than a 7B I tried a week ago. Maybe it's the

<!-- SC_OFF --><div class="md"><p>I'm new to the local hosting, and I have just tried 2B models on my smartphone (qwen2.5/3.5, gemma). </p> <p>I have asked generic questions, like

<!-- SC_OFF --><div class="md"><p>Question: can i do the thinking with a smaller model, like Gemma 4 4B, then use that as the prompt for Gemma 4 31B, to speed things up?</p>