Llama.cpp on a jailbroken PS5: run any GGUF model natively on the console github.com

Developer Cobanov’s PS5LM is a homebrew app for jailbroken PS5 consoles that runs GGUF models with llama.cpp on the console’s GPU and serves them over a local network through OpenAI- and Anthropic-compatible APIs. On a PS5 Slim, it reports speeds of 55 tokens per second for a 9B model and about 20 tokens per second for a 27B model; it also supports decision models and image input.