There is a lot of room for improvement still. Fable or GPT 5.6 Sol regularly hand off obviously broken features in my mobile app. QA is the bottleneck. If vi

There is a lot of room for improvement still. Fable or GPT 5.6 Sol regularly hand off obviously broken features in my mobile app. QA is the bottleneck. If vision capabilities and understanding of motion improve significantly, perhaps the models can tell that a list is not scrolling correctly, or a transition not animating as it should.