Jailbroken PS5 runs Qwen 3.8 27B on its GPU at 20 tok/s with PS5LM Plus
98Saman · reddit · 2026-10-15
Someone jailbroke a PS5 and got llama.cpp running natively on its GPU. PS5LM Plus, a fork of cobanov's PS5LM, runs GGUF models directly on the console with a TV dashboard, a model store, and an OpenAI-compatible API for every device on your network. The dashboard shows Qwen 3.8 27B (UD-IQ2XXS, 6.8 GiB) generating at 20.9 tok/s, 86W power draw, 91% memory usage, 55°C SoC, and 4.1 J/token. A game console turned local LLM server — a delightfully hardcore hack.
More from Infra
- Neocloud stocks fall 12% in a day after CoreWeave's bond selloff — SumitGup · 2026-10-15
- TRELLIS.2 image-to-3D runs at full 1536 resolution on a 16 GB card via tiling patches — over-yonder-tech · 2026-10-15
- ServeTheHome tours AWS's secretive networking lab: timing, optics and hyperscale — firstadopter · 2026-10-15
- vLLM adaptive verification now works head-free, using draft logits with online logistic calibration — vllm_project · 2026-10-15
- One R9700 running Qwen3.6-35B did $426.80 of equivalent Opus coding work in a week — supracode · 2026-10-15
- Eric Jang's three signals of AI overinvestment: compute repricing, open-source catch-up, robot cost halving — ericjang11 · 2026-10-15