llama.cpp on a jailbroken PS5: run any GGUF model natively on the console, starting with Qwen 3.8
Release v0.1.1 in the Payload Manager source; close out Phase 1.5
Mert Cobanov256050fc3d15
diff --git a/docs/ROADMAP.md b/docs/ROADMAP.mdindex 530a03e..c407307 100644--- a/docs/ROADMAP.md+++ b/docs/ROADMAP.md@@ -61,7 +61,7 @@ An Ollama-like way to get models and chat, without a PC in the loop. One payload- [x] **Run**: llama-server runs inside the same process (`llama_server()`), and switching models stops it (`llama_server_terminate()`) and starts it again, with the model locked in memory- [x] **Chat** in a plain page (llama.cpp's own UI renders blank in the PS5 browser), on the TV or from any device on the network, and the OpenAI-compatible API on port 8081-**Done when** a model can be picked, downloaded and chatted with using only the DualSense.+**Done when** a model can be picked, downloaded and chatted with using only the DualSense. Done in v0.1.0 (2026-10-02); v0.1.1 restyled the pages. Payload Manager installs it from `https://ps5lm.cobanov.dev/payloads.json`, and a listing in the default mirror is requested (itsPLK/ps5-payloads-mirror#19).## Phase 2: fit and speed on the CPUdiff --git a/scripts/update-source.sh b/scripts/update-source.shindex 1a7ee7b..bcf33bd 100755--- a/scripts/update-source.sh+++ b/scripts/update-source.sh@@ -10,15 +10,14 @@ set -euo pipefailroot="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")/.." && pwd)"version=${1:?usage: update-source.sh VERSION}url="https://github.com/cobanov/PS5LM/releases/download/$version/ps5lm.elf"-tmp=$(mktemp)-trap 'rm -f "$tmp"' EXIT-# Hash what the release actually serves, not a local build.-curl -fsSL -o "$tmp" "$url"-if command -v sha256sum >/dev/null; then- sum=$(sha256sum "$tmp" | cut -d' ' -f1)-else- sum=$(shasum -a 256 "$tmp" | cut -d' ' -f1)+# The checksum of the asset as GitHub records it. Downloading it to hash+# can return a cached older file when a release was recreated.+sum=$(gh api "repos/cobanov/PS5LM/releases/tags/$version" \+ --jq '.assets[] | select(.name == "ps5lm.elf") | .digest' | sed 's/^sha256://')+if [ -z "$sum" ]; then+ echo "update-source: no ps5lm.elf in release $version" >&2+ exit 1fipython3 - "$version" "$url" "$sum" "$root/site/payloads.json" <<'EOF'diff --git a/site/index.html b/site/index.htmlindex 915a875..91cbbe4 100644--- a/site/index.html+++ b/site/index.html@@ -42,7 +42,7 @@<h1 class="display">llama.cpp, running on a PlayStation 5.</h1><p class="lede">PS5LM runs open language models on the console's own CPU. You chat with them in the PS5's browser, typing with the DualSense.</p><div class="cta">- <a class="btn btn-primary" href="https://github.com/cobanov/PS5LM/releases/latest">Download v0.1</a>+ <a class="btn btn-primary" href="https://github.com/cobanov/PS5LM/releases/latest">Download PS5LM</a><a class="btn" href="https://github.com/cobanov/PS5LM">Source on GitHub</a></div><p class="hero-note">One payload, <code>ps5lm.elf</code>: pick a model, it downloads on the console, chat. Open source under GPL-3.0, for a PS5 you own on firmware 13.60 or lower.</p>diff --git a/site/payloads.json b/site/payloads.jsonindex 2e28daa..08cf970 100644--- a/site/payloads.json+++ b/site/payloads.json@@ -1,14 +1,14 @@[{"name": "PS5LM",- "filename": "PS5LM_v0.1.0.elf",- "url": "https://github.com/cobanov/PS5LM/releases/download/v0.1.0/ps5lm.elf",+ "filename": "PS5LM_v0.1.1.elf",+ "url": "https://github.com/cobanov/PS5LM/releases/download/v0.1.1/ps5lm.elf","source": "https://github.com/cobanov/PS5LM/releases",- "source_direct": "https://github.com/cobanov/PS5LM/releases/download/v0.1.0/ps5lm.elf",+ "source_direct": "https://github.com/cobanov/PS5LM/releases/download/v0.1.1/ps5lm.elf","description": "Run local LLMs (llama.cpp) on your PS5: download a model on the console and chat in the browser","last_update": "2026-10-02",- "version": "v0.1.0",+ "version": "v0.1.1","category": "Utilities & Tools",- "checksum": "26d7a543bec9c9ff0ccdb1fa067252254ccce88b81e5c293c22c1f536d0cfd5e"+ "checksum": "10ee1a913e2d534fc81e1822e5969a6d70740ff88f4eb7c8fe98789d22152126"}]