Gemini and antigravity are underrated

At the risk of coming across as a Google shill, this is my hot take. When it comes to smaller coding tasks like iterating on front end design or fixing smaller bugs, the experience of using Gemini flash in Antigravity is WAY better than GPT 5.6 in codex. I'm mostly comparing Flash 3.5 High to 5.6 Luna (across the effort spectrum), though I'd say flash is better for this work than terra or sol as well. The sheer speed and tasteful constraint of flash makes it excellent to go back and forth with when building. It might take more steering to finish a problem, but you'll end up with a product much closer to your initial vision in less time if you get good with this workflow. The main problem with GPT 5.6 is the slow speed and ridiculous tendency to over-engineer every single feature you ask for. It has a higher ceiling of intelligence, but even Luna at lower effort levels can't help itself from overthinking all the time. Sol is great to throw at a problem too big foot gemini flash to handle, but for everything else, i find myself going to antigravity. If I send 5.6 on smaller tasks, the output is slower, and I end up having to do several iterations removing all the unnecessary complexity it shoved in without me asking for it. Now with the launch of Flash 3.6, I'm actually feeling quite optimistic for Google. I know it looks unimpressive on benchmarks, but seriously give it a try yourself on some of your more simple day to day tasks. The speed and simplicity is so underrated. Also I think Antigravity's usage limits are more generous if you don't count the free codex resets. I'm mostly working on fairly simple game dev and web development, so maybe my perspective is skewed because most people's tasks are beyond this. But if you haven't given flash a fair shot, take this as your sign to try and let me know.