I don't think it will work out. Brothers, you might as well go with small language models (SLM) instead of large language models (LLM). For example, if you're building an AI Agent for auto-replying to emails, using a large model is like having a professor tighten screws—you'd be better off hiring a professional organizer.

Large models are like elite, overseas-educated nannies. They can do everything—tutor kids in advanced math, have full English conversations, prepare elaborate multi-cuisine feasts, and they're only 24 years old fresh out of college. The downside? They cost 50k per month, and every time you ask them to take out the trash, they need 3 minutes to contemplate the meaning of life.

You might as well just use a small model instead. It'll clean your windows every day for just 500 per month. Regular people can't solve massive commercial coding problems anyway, but right now there's a real gap in the market for specialized small models.