Jailbroken PS5 runs llama.cpp natively, serving any GGUF model over your network

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

PS5LM is a homebrew app that runs llama.cpp on a jailbroken PS5's RDNA 2 GPU, using 11.4 GB of shared memory. It serves OpenAI and Anthropic APIs, streams every request to the TV, and charts console temperature and load. Qwen 3.5 9B hits 55 tokens per second; Qwen 3.8 27B runs at about 20 tokens per second with 64k context. It also supports decision models via Ollaya's /v1/systemone endpoint.