{"slug": "ollada-open-diffusion-models-locally-with-numbers-you-can-trust", "title": "Ollada: open diffusion models, locally, with numbers you can trust", "summary": "A developer released Ollada, an open-source tool that runs open diffusion language models locally with a single command, bundling five models (DiffCoder-8B, Dream-7B, SDAR-4B, SDAR-1.7B, and NVIDIA's Nemotron-3B) behind short names. Each model is listed only after being run and measured, with VRAM figures such as DiffCoder at 13.65 GB, Dream at 9.13 GB, and SDAR-4B at 8.34 GB on a 16GB RTX 5060 Ti. The SDAR and Nemotron models also run on a laptop CPU, though the developer notes limits including no token streaming yet and slow CPU runs on long answers.", "body_md": "Ollada - run open diffusion language models on your own computer.\n\n[https://opensourcerulesz.github.io/ollada](https://opensourcerulesz.github.io/ollada)\n\n[https://github.com/OpenSourceRulesz/ollada](https://github.com/OpenSourceRulesz/ollada)\n\nMost chatbots write word by word. A diffusion language model writes a whole answer at once, then fixes it over a few passes.\n\nThe problem is those models are scattered across Hugging Face, each with different, incompatible code, and nobody tells you if one will actually run on your machine. Ollada fixes that: one shelf, short names, one command, and every model is only listed after we've actually run it and measured it.\n\nTry it, no GPU needed:\n\npip install \"ollada[cpu]\"\n\nollada pull sdar:1.7b\n\nollada run sdar:1.7b\n\nIf you get stuck, step-by-step setup instructions are on the site.\n\nFive models: diffcoder:8b for coding, dream:7b, sdar:4b, sdar:1.7b, and nemotron:3b from NVIDIA. On a 16GB RTX 5060 Ti we measured DiffCoder 13.65 GB, Dream 9.13 GB, SDAR-4B 8.34 GB. On a GPU, start with sdar:4b. The SDAR models and Nemotron also run on a laptop CPU, slower but no GPU required.\n\nHonest limits: no token streaming yet, CPU runs are slow on long answers, Dream is eh\n\nFree, local, $0. Weights come from Hugging Face\n\nI’m open to any conversations !", "url": "https://wpnews.pro/news/ollada-open-diffusion-models-locally-with-numbers-you-can-trust", "canonical_source": "https://dev.to/opensourcerulesz/ollada-open-diffusion-models-locally-with-numbers-you-can-trust-3i78", "published_at": "2026-10-29 13:35:37+00:00", "updated_at": "2026-10-29 13:47:32.878573+00:00", "lang": "en", "topics": ["large-language-models", "ai-tools", "developer-tools", "ai-infrastructure"], "entities": ["Ollada", "Hugging Face", "NVIDIA", "DiffCoder", "Dream", "SDAR", "Nemotron"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/ollada-open-diffusion-models-locally-with-numbers-you-can-trust", "markdown": "https://wpnews.pro/news/ollada-open-diffusion-models-locally-with-numbers-you-can-trust.md", "text": "https://wpnews.pro/news/ollada-open-diffusion-models-locally-with-numbers-you-can-trust.txt", "jsonld": "https://wpnews.pro/news/ollada-open-diffusion-models-locally-with-numbers-you-can-trust.jsonld"}}