Everyone is excited about GPT-OSS 20B and the prospect of running this LLM on their phones. However, the iPhone 16 only has 8 GB of RAM, so it isn't possible to run the model locally. I will be surprised if the next iPhone could do that.