Qwen 3.8 Next Flash appreciation post

Article automatically generated from technical news.

I don't want to talk about the performance and technical things, but about how I work with my hobby projects has changed thanks to this model. My machine has generated around 25M tokens since the model came out, and I've done a mental retro on it. I've found it to have incredible prompt adherence. I can leave to run it by itself and get back to it, and find that it did exactly what I've asked it to. I've only ever seen it go astray once, where I've asked something that is too high level a