![](/static/253f0d9b/assets/icons/icon-96x96.png)
![](https://lemmy.ml/pictrs/image/d3d059e3-fa3d-45af-ac93-ac894beba378.png)
What language are you programming in? In swift I have found all models (including sonnet) next to useless. Tells me something wrong almost every question i ask, has made up macros and apis, etc.
For English I have found Claude models slightly better than the GPT 4 subscription I used to have. For anything in multiple (human, not programming) languages, gpt has seemed best for me.
I thought they had confirmed at least some of the image generation stuff happening locally. I am in the intelligence beta now and went offline and played around with Siri and lots of stuff worked. Not really doing much new right now, but the speed and quality of understanding and dealing with when you stumble over words are way better.