A Reddit user seeks recommendations for optimal local AI models for coding and UI/UX design on a Ryzen AI Max+ 395 system with 128GB unified memory. The developer is migrating from OpenAI Codex due to usage limits and costs, and has initially tested Hermes with Qwen3.8-Flash-Next-GSQ-RCO. The query targets the LocalLLM community for hardware-specific model optimization advice.