Interesting, I haven't seen it actually return nonsense yet. (Some incorrect things and getting into thinking loops, but always coherent) I'm running it on a latest llama.cpp with the bf16 gguf. What are you using?
Overall if you're memory constrained it's probably still worth to try and fiddle around with it if you can get it to work. Speedwise if you got the memory a 5090 can get ~50-100tok/s for a single query with 32B-AWQ and way more if you have something parallel like open-webui