And I’m absolutely loving it’s behaviour! Currently using it in Kilo and it’s a good model. It’s not that fast yet, but that’s mainly a provider issue as you can see on OpenRouter:

Most of the providers haven’t allocated enough resources or haven’t optimised it enough, that’s an assumption but what I usually see is that the throughput goes up in the days after release. It’s also many people trying out the model. It’s already the #10 for programming on OpenRouter.

The model debugged one of the more complex issues I had in one-shot. The only instruction I gave the model was an exception trace which showed that some things never populated the database and it did it with the cost of about $1.

So in my option, that’s kinda awesome. The cool thing was that it, like other models, just ran it’s own work to see if the issue was gone.

If you want to try out Kilo yourself, use my link here which helps a lot!