{"slug": "opencv-5-0-released-with-rewritten-dnn-engine-built-in-llm-vlm-support", "title": "OpenCV 5.0 Released With Rewritten DNN Engine, Built-In LLM & VLM Support", "summary": "OpenCV 5.0 was released today as a major update to the open-source computer vision library, featuring a rewritten deep neural network engine and built-in support for large language and vision language models. The new version also includes ONNX coverage exceeding 80%, a hardware abstraction layer, and an improved 3D vision toolkit, with optimized paths for Intel, Arm, Qualcomm, and RISC-V processors. Developers plan to add native GPU support to the DNN engine in future updates.", "body_md": "# OpenCV 5.0 Released With Rewritten DNN Engine, Built-In LLM & VLM Support\n\nOpenCV 5.0 released today as a major update to this widely-used, open-source computer vision (CV) library.\n\nThe OpenCV library is widely-used for ral-time computer vision needs and machine learning applications. With OpenCV 5.0, it's a huge release and a major step forward in advancing its already leading capabilities.\n\nOpenCV 5.0 features a rewritten deep neural network (DNN) engine, ONNX coverage surpassing 80%, built-in large language model (LLM) and vision language model (VLM) support, and a new hardware abstraction layer as well as a much better 3D vision toolkit.\n\nOpenCV 5.0 currently has tuned paths for Intel IPP with SSE/AVX-optimized kernels, Arm KleidiCV, Qualcomm FastCV, and RISC-V Vector RVV. Coming up next the OpenCV developers plan to work on native GPU support within their new DNN engine.\n\nOpenCV 5.0 is performing very well against the Microsoft ONNX Runtime:\n\nMore details on the huge release of OpenCV 5.0 via today's announcement on\n\nThe OpenCV library is widely-used for ral-time computer vision needs and machine learning applications. With OpenCV 5.0, it's a huge release and a major step forward in advancing its already leading capabilities.\n\nOpenCV 5.0 features a rewritten deep neural network (DNN) engine, ONNX coverage surpassing 80%, built-in large language model (LLM) and vision language model (VLM) support, and a new hardware abstraction layer as well as a much better 3D vision toolkit.\n\nOpenCV 5.0 currently has tuned paths for Intel IPP with SSE/AVX-optimized kernels, Arm KleidiCV, Qualcomm FastCV, and RISC-V Vector RVV. Coming up next the OpenCV developers plan to work on native GPU support within their new DNN engine.\n\nOpenCV 5.0 is performing very well against the Microsoft ONNX Runtime:\n\nMore details on the huge release of OpenCV 5.0 via today's announcement on", "url": "https://wpnews.pro/news/opencv-5-0-released-with-rewritten-dnn-engine-built-in-llm-vlm-support", "canonical_source": "https://www.phoronix.com/news/OpenCV-5.0-Released", "published_at": "2026-06-06 12:06:13+00:00", "updated_at": "2026-06-06 12:28:47.713474+00:00", "lang": "en", "topics": ["computer-vision", "machine-learning", "large-language-models", "neural-networks", "artificial-intelligence"], "entities": ["OpenCV", "Intel", "Arm", "Qualcomm", "RISC-V", "ONNX", "Microsoft"], "alternates": {"html": "https://wpnews.pro/news/opencv-5-0-released-with-rewritten-dnn-engine-built-in-llm-vlm-support", "markdown": "https://wpnews.pro/news/opencv-5-0-released-with-rewritten-dnn-engine-built-in-llm-vlm-support.md", "text": "https://wpnews.pro/news/opencv-5-0-released-with-rewritten-dnn-engine-built-in-llm-vlm-support.txt", "jsonld": "https://wpnews.pro/news/opencv-5-0-released-with-rewritten-dnn-engine-built-in-llm-vlm-support.jsonld"}}