After my experience with vLLM on NVIDIA hardware, I was painfully aware how behind Apple Silicon was in terms of software support. My 11 tokens per second on MacOS versus the 30 per second I saw on...

Last time I wrote about pushing my M4 Pro to its limits and teasing what might come next. That next thing arrived: a Dell Pro Max GB10 I could use for a month, courtesy of Willem. It is basically a...

After weeks of experimenting with running AI models locally on my MacBook Pro I can share some more of my experiences and insights. I have enjoyed my setup a lot! I could for example keep working i...

So you’re thinking about running AI locally? Welcome to the club! After spending many hours tinkering with different models, dealing with mysterious errors, and watching my laptop fan work overtime...