AIXDAabout 16 hours ago
Three tiny local LLMs survive testing
1 min read
Nolen Jonker tested sub-4B local LLMs and kept only three: Google's Gemma 4 E2B, Alibaba's Qwen 3.5 small series, and Qwen 2.5 Coder 3B Instruct. Gemma 4 E2B runs on phones and Chromebooks with 128K context and multimodal input. Qwen 3.5 offers 262K native context and tool calling. Qwen 2.5 Coder handles 92 programming languages at 1.9GB.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
