I’m a web developer doing mostly coding, but also project management, requirements analysis, testing, etc. I recently started experimenting with local LLMs, mostly because agentic stuff finally made them feel useful. Note: This text was fed to chartgpt to fix my messy repeating grammar My initial impression was honestly pretty discouraging. Endless model option confusion, benchmarks that are hard to translate, huge VRAM requirements and hardware prices that are completely unreasonable. Still...