{"slug": "llama-cpp", "title": "llama.cpp", "summary": "Llama.cpp, the open-source project that runs AI models locally on consumer hardware, has launched a new installer and command-line tool at llama.app, enabling users to run frontier AI models such as Alibaba's Qwen 3.6, Google's Gemma 4 and Gemma 3, and OpenAI's GPT-OSS entirely on their own machines with no API keys or telemetry. The tool pairs with the local coding agent Pi via the pi-llama plugin, automatically discovering local models and keeping files and requests on-device, and is optimized for any hardware from laptops to clusters.", "body_md": "# AI that lives on your computer. Open-source, private & always local.\n\nRun frontier AI entirely on your machine. No API keys, no telemetry, no limits. Own your models and conversation data.\n\n`curl -LsSf https://llama.app/install.sh | sh`\n\n## Pair it with a local coding agent.\n\nRun `llama serve`\n\n, install the `pi-llama`\n\nplugin and launch [Pi](https://github.com/earendil-works/pi). It will automatically discover your local model. No config, no API keys. Files stay on your machine, requests never leave it.\n\n```\n# 1. Serve a model\nllama serve\n\n# 2. Install the pi-llama plugin\npi install git:github.com/huggingface/pi-llama\n\n# 3. Run Pi, everything is set\npi\n```\n\n## Optimized for any hardware.\n\nFrom your laptop to a cluster, llama.cpp runs on whatever you have. Same binary, same models, same hand-tuned kernels for every GPU and CPU.\n\n## Run your first model\n\n### Qwen 3.6\n\nAlibaba's next-gen natively multimodal reasoning models. Dense and MoE variants that rival models many times their size on coding and vision tasks.\n\n### Gemma 4\n\nGoogle's most capable open models, built from Gemini 3 technology. Supports multimodal reasoning, agentic workflows, and 140+ languages.\n\n### GPT-OSS\n\nOpenAI's first open-weight models since GPT-2. Built for reasoning, agentic tasks, and developer use with function calling and tool use capabilities.\n\n### Gemma 3\n\nGoogle's multimodal models built from Gemini technology. Supports 140+ languages, vision, and text tasks with up to 128K context for edge to cloud deployment.", "url": "https://wpnews.pro/news/llama-cpp", "canonical_source": "https://llama.app", "published_at": "2026-08-12 04:51:59+00:00", "updated_at": "2026-08-12 05:11:22.005723+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-tools", "developer-tools"], "entities": ["llama.cpp", "Alibaba", "Google", "OpenAI", "Pi", "pi-llama", "Qwen 3.6", "Gemma 4"], "alternates": {"html": "https://wpnews.pro/news/llama-cpp", "markdown": "https://wpnews.pro/news/llama-cpp.md", "text": "https://wpnews.pro/news/llama-cpp.txt", "jsonld": "https://wpnews.pro/news/llama-cpp.jsonld"}}