45 security apps cut the ram and CPU performance in half out of the gate, then you have Win11 with its web gui and users who complain about any slowness so i7 or r7 is about the norm most places / bulk big 3 oems make for business.
personally I have an i7 in the lab that turned into a crawl on Youtube because they started sending VP9 video by default, but i7 doesn't have hardware decode for it. Got a plugin to force h264 (h264ify) and it's back to normal.
Then you have situations where a software update adds a CPU instruction dependency and straight up doesn't work anymore.
Had to install and force-retain an old version of Spotify client on a bench lab PC because the new versions require AVX-1/2. It's a streaming audio player...
What local model would even come close? Kind of feels like you're not using/used SOTA models if they're realistic alternatives to the same kind of tasks. Qwen-3.8-27B-Abliterated-by-MaxxedWeightsGuy82 or similar isn't gonna cut it, almost certainly.
GLM 5.3 is probably the best open weight model for cybersecurity/exploit development right now. Though it is still significantly behind the proprietary ones and you probably need your own datacenter to run it effectively. Same goes for the full Qwen 3.8 model. You can try the smaller versions, but even more capability will get left on the table that way.
I run an abliterated distillation of Qwen 3.8 27B, slightly quantized to fit on my 4090, and I've been evaluating it to use as a worker bee for research directed by a smarter model.
Much like in the article, abliterated Qwen will not obey restrictions on its behavior encoded in the prompt. If you want something not to happen, it better be enforced in the harness or environment (e.g. sandbox). It is much different than the Anthropic models I'm used to, which will, the vast majority of time, follow rules (before auto mode, I used to always run them in "yolo" mode).
I am curious whether there's a connection between abliteration and rule following. These abliterated models are the ones you most want to follow your rules.
I would say in this case abliteration is the likely culprit. To uncensor a model this way, you literally deactivate the parts that would enact refusals. As in things it was told not to do. But the real process is more like brain surgery performed by a alchemist according to an ancient religious book where noone involved really understands what is actually happening in the model.
I noticed your joy-check link 404's now... I tried poking around your /skills/ folder but didn't find it easily. Should you still have that available I'd love to check it out.
You'll need tons of tokens for these to churn through heavy reasoning and feedback driven tasks. You might even want to run many agents in parallel. Combine that with GPU memory requirements in the terabyte range just to hold the weights and you've pretty much left local consumer setups entirely.
Note that Mac Studio Ultra M3 (or upcoming M5) with 512GB is effective. You don't have to do this work fast, overnight is fine.
Unless trying to use it interactively and adversarially, in which case it's not fast enough plus would be why those of us without our own datacenters will get told we can't have nice things.
This month I reduced my spend on cloud AI and started using local models as primary - it's maybe 2 or 4 times slower for some things but it still is getting the job done just fine. I talked to a friend of mine who's been doing the same running smaller Qwen models locally for coding.
I'm going to guess that the agents are built this way on purpose. I just finished watching BlackBerry and Flash of Genius and yeah this is American business ethics just operating as normal.
Cellpose[0] and Stardist[1] are the two you'll see most heavily run and talked about on the forums[2]. They're classic CNN's, but it just feels like they (and any modern models) are swept up in "Go out and buy a 4090.But these guys will give you plenty of mileage before you have to reach for the bigger ones.
reply