I'm particularly excited about local LLM based line completion, as I've found their python equivalent to be really helpful (particularly as someone that doesn't do a huge amount of python)

After using this for a few hours today, my review is mixed. When it gets it right, it's a huge productivity booster, particularly writing out predictable unit tests, or mapping new pieces of data through several layers of code.

I did experience quite a few ide lockups though, particularly when doing cut / paste operations, and it also got things wrong pretty often where the traditional auto complete probably would've worked better.

On balance I'll keep using it, and hope that it gets continued efforts to improve it over the next few releases.

I also hope that it remains fully local, and if it isn't already starts taking advantage of my beefy Nvidia GPU - I didn't check but the freezes make me suspect it might be running on CPU