Eh. Apple's chips are good but not exactly game changing. At the end of the day if they're doing anything useful they still run the same software everyone else'

Eh. Apple's chips are good but not exactly game changing. At the end of the day if they're doing anything useful they still run the same software everyone else's do with roughly the same performance. It's a great analogy for the situation with language models. Can you get a slight edge with the particular RL tuning on some models? Sure. Does it actually result in qualitative changes in what the model can do? No not really, all of them can do MCP, all of them still hallucinate etc.