The Reddit post questions why the AI community focuses heavily on building compact coding models when users also need strong world knowledge to replace large providers. It notes interest in sub‑50 GB agentic and tool‑calling models but argues that better world knowledge is essential for personal AI use. The author highlights Qwen 3.8 27B as a promising step toward balancing coding ability with broader knowledge.
Read original
reddit/r/LocalLLaMA