Can a 4B local model actually feel like an AI assistant?
DGX agentI've been building Arcon around Qwen3-4B + LoRA. Instead of just making it a chatbot, I'm experimenting with persistent memory, personality/mood, internal state, tools, and eventually having it proces