Qwen 3.8 Flash Next This model should be trained in a hurry
Qwen 3.8 Flash Next This model should be trained in a hurry, so this model cannot run uncensored models. Running uncensored models with long context will cause garbled code. Yesterday I transplanted the original Flash Next again. The speed is pretty good and there should be no problem with stability. And the speed is much faster, probably up to 69 TPS for a single stream…
Looked into local ML video upscalers for anime.
Qwen-3.8 did the research, the setup of the top 3 options, the qualitative work of choosing the best one, etc.
I gave Qwen 3.8 Max two hours and a hard problem to see what would break.
The task: build a working iPhone Duo in the browser. Apple published the Duo as an AR asset before the device shipped, so the real geometry is out there. Load it, fold it, and put a live website on the
I’m a lawyer from Spain, not a professional developer
But over the last few months I’ve been using AI agents to build projects that I genuinely wouldn’t have known how to approach before. This is CMMChat, my own AI client and one of the projects I’m currently building with
You have reached the end of the archive
All of qwen38