The project's primary site should not be GitHub, and I am serious. Playstation 5 is not some historical curiosity and Sony might strike the page down any hour now.
Those projects should default to hosts less geared toward corporate governancy; custom internet domain being the first requirement.
I'm sure Sony has an opinion of the jailbreaks that enable this, but they really don't give a shit about anyone running llama.cpp. Why would they? They're just glad you're not pirating games.
Is there a real or imagined hardware subsidy on the PS5 hardware though? In other words, is the current MSRP below what Sony would charge in a hypothetical world where they weren’t involved in any form of gaming software business, with an assumption that they will make additional margin dollars from game sales? I can see some manager making the case that repurposing the hardware for non-gaming use is taking food off his family’s table, in a manner of speaking. If so, it’s an argument that you don’t want this type of use to become popular and have PS5s start flying off shelves and never having games bought for them.
There's no feasible way to avoid takedowns/DMCA in 2026 unless you do TOR shenanigans. Even with a custom domain, Sony could just DMCA the web host for the project.
Sure you can. This repo has definitely been vacuumed into the latest training data dump, so the next generation of models will happily spit out a similar reproduction, Sony be damned!
Torrents. Got the leaked GTA6 videos by torrenting and they're still easily available on public trackers. Don't see why projects like these couldn't be distributed similarly if shit hits the fan.
Even though it only runs qwen 3.8 27b at a 2 bit quant, it's still highly viable, especially for things where you just need a constant token stream of slightly below average intelligence, ie this would be great for things like:
- monitoring server error rates
- reviewing security footage
- watching plant growth / controlling watering
etc. Any long running task you can now just slap a PS5 on it. 70 cents a day in power cost means you now get ~2M output tokens per day (~$3.50 worth of usage). It's not the worst thing in the world if you have one lying around. 2bit isn't even that terrible - it's ~80% the accuracy of 27b at full quant, which should put it at around GPT-5 levels of intelligence (which honestly is crazy in its own right).
I'm sure someone will point out some spec that's better on the PS5, but the extra $200 will make up the difference if you spend more time looking than the 15 seconds I just did.
>I'm sure someone will point out some spec that's better on the PS5
They're not even in the same ballpark - the PS5 is more than an order of magnitude better in GPU performance.
Despite how much it does matter, RAM is not the only thing you need for running LLMs - that N150 is not going to run anything but really tiny models at any sort of reasonable speed - certainly nothing that approaches using all of its ram. The GPU in the PS5 is ~25x faster. And if you're running games the RAM doesn't really affect performance at all once you have "enough" for a given game.
You're right that you probably could build something quite a bit faster than a PS5 for the same price though - if you can build/find everything but the GPU for ~250 USD and slap a 16gb 9060xt (~25tflops @32bit compared to the PS5 gpus ~10tflops @32bit) in it for another ~350 USD it should theoretically do a better job - the GPU in the PS5 was on par with mid-high end PC GPUs on launch, but that was like 5 years ago.
N150 isn't really what you'd want to use for a gaming machine. At a budget of mid $300s - the best (new) Mini PC you're going to find right now is probably this unit[1] with a Ryzen 7 7735U (680M iGPU) and 16GB RAM. Not the most powerful device - but I have a similar one I use to play emulator games, indie games on steam - throw Steam OS on it and it's a great living room device. Expand that budget to $530 and you can get this guy[2] with a Ryzen 7 255 (780M iGPU) - which will play modern games (on low settings for many).
Conceivably this should also work on the BC250 cards, which are ex-cryptocurrency mining motherboard built around AMD's "Cyan Skillfish" APU which is a cut-down version of the PlayStation 5's custom silicon. You can still pick them up on ebay for $150-$200.
I paid $150 shipped a few months ago then maybe $40-50 for all the parts to run it. The RTX 3060 sells for $400+ in my market and still needs a computer to attach it to!
Flashing the bios takes ten minutes with a thumb drive.
It’s hard to get wrong but guess some people would be nervous enough about that to pay $20 extra.
With this in mind, having LLM-powered NPCs would be a beautiful thing. There's been limited testing on PC games (PUBG Ally), but I can imagine it will be the standard.
A 20 tps Qwen3.8-27B at 2 bits quantization is limited but not useless. I've been putting one on "read this codebase and figure out XYZ" for me, where the typos are less consequential.
For somebody who hasn't kept up with all the different open source models and what token rates mean, how would using this compare to something like codex or claude? I assume it would be much slower and less capable..?
Those projects should default to hosts less geared toward corporate governancy; custom internet domain being the first requirement.
I struggle to think of anywhere on the entire open internet you could put something like this where it could not be taken down.
- monitoring server error rates
- reviewing security footage
- watching plant growth / controlling watering
etc. Any long running task you can now just slap a PS5 on it. 70 cents a day in power cost means you now get ~2M output tokens per day (~$3.50 worth of usage). It's not the worst thing in the world if you have one lying around. 2bit isn't even that terrible - it's ~80% the accuracy of 27b at full quant, which should put it at around GPT-5 levels of intelligence (which honestly is crazy in its own right).
I'm sure someone will point out some spec that's better on the PS5, but the extra $200 will make up the difference if you spend more time looking than the 15 seconds I just did.
(or if you just build a PC from parts yourself)
They're not even in the same ballpark - the PS5 is more than an order of magnitude better in GPU performance.
Despite how much it does matter, RAM is not the only thing you need for running LLMs - that N150 is not going to run anything but really tiny models at any sort of reasonable speed - certainly nothing that approaches using all of its ram. The GPU in the PS5 is ~25x faster. And if you're running games the RAM doesn't really affect performance at all once you have "enough" for a given game.
You're right that you probably could build something quite a bit faster than a PS5 for the same price though - if you can build/find everything but the GPU for ~250 USD and slap a 16gb 9060xt (~25tflops @32bit compared to the PS5 gpus ~10tflops @32bit) in it for another ~350 USD it should theoretically do a better job - the GPU in the PS5 was on par with mid-high end PC GPUs on launch, but that was like 5 years ago.
[1] https://www.amazon.com/dp/B0GGM6BKN9 [2] https://www.amazon.com/dp/B0FX2TV3HX
Shameless plug - I methodically rank the best possible Mini PC deals on https://MiniPCs.zip for this exact purpose.
And at that price point or a little bit more you might as well just get an RTX 3060 and save yourself a lot of hassle (you also get DLSS, DLAA etc.)
The readme oversells it by pretending it would be something actually usable, when instead it is a cool achievement to make it happen.
https://www.ncsa.illinois.edu/2003/05/27/playing-the-superco...
However you can run comparable or better models than gpt4. Try to stay under 30b