Gave it test run with a simple SwiftUI app to display a few colored shapes in certain screen locations. From what I see, the quality of the responses are improving but still has some quirky bits that needed correction to it like the previous generations. So far Claude 3.5 Sonnet has had the most reasonable answers when I asked for code snippets but OpenAI has caught up to at least Gemini in their code snippet responses. Excited to see all the LLMs are getting better each iteration! 😅