@ts_floydThe question worth asking about local models isn't whether they're as smart as the big ones. It's which of my 1000 calls ever needed the smart one. Andrej Karpathy puts it plainly: people are renting frontier models for jobs a local 3B model could do, and Small Local LLM's are the future. His full argument for moving that core onto your own machine runs 18 pages, laid out as Local Language Model Engineering 2026, with a Cognitive Core Thesis and Ten Steps From One Command to an Always-On Stack.

@ts_floydThe question worth asking about local models isn't whether they're as smart as the big ones. It's which of my 1000 calls ever needed the smart one. Andrej Karpathy puts it plainly: people are renting frontier models for jobs a local 3B model could do, and Small Local LLM's are the future. His full argument for moving that core onto your own machine runs 18 pages, laid out as Local Language Model Engineering 2026, with a Cognitive Core Thesis and Ten Steps From One Command to an Always-On Stack.