The open source models are pretty good too now. They are a few months behind, but not more than that. Sure, you still have to host them in the cloud to get enou

The open source models are pretty good too now. They are a few months behind, but not more than that. Sure, you still have to host them in the cloud to get enough VRAM to run them - but looking 10+ years into the future the end game here will probably be having a local LLM that runs on your own computer and is more than capable enough to do coding for you.