--- title: New DNN engine with enhanced ONNX and LLM/VLM support,... description: OpenCV 5 has been released with a rebuilt DNN engine that expands ONNX operator coverage from ~22% to over 80%, adds LLM/VLM support (Qwen 2.5, Gemma 3,... canonical: https://daily.dev/posts/new-dnn-engine-with-enhanced-onnx-and-llm-vlm-support-intel-arm-and-risc-v-hardware-optimizations-5jrhxdsuq twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: New DNN engine with enhanced ONNX and LLM/VLM support, Intel, Arm, and RISC-V hardware optimizations | daily.dev og:description: OpenCV 5 has been released with a rebuilt DNN engine that expands ONNX operator coverage from ~22% to over 80%, adds LLM/VLM support (Qwen 2.5, Gemma 3,... og:url: https://daily.dev/posts/new-dnn-engine-with-enhanced-onnx-and-llm-vlm-support-intel-arm-and-risc-v-hardware-optimizations-5jrhxdsuq og:image: https://api.daily.dev/og/posts/5JrHXDSUQ.png og:image:alt: New DNN engine with enhanced ONNX and LLM/VLM support, Intel, Arm, and RISC-V hardware optimizations og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # New DNN engine with enhanced ONNX and LLM/VLM support, Intel, Arm, and RISC-V hardware optimizations **[CNX Software](https://daily.dev/sources/cnx-software)** · [@cnxsoft](https://daily.dev/cnxsoft) · 5 min read · 0 upvotes · 0 comments ## Summary OpenCV 5 has been released with a rebuilt DNN engine that expands ONNX operator coverage from ~22% to over 80%, adds LLM/VLM support (Qwen 2.5, Gemma 3, PaliGemma, GPT-2/4 family) via a native tokenizer and KV-cache, and introduces four selectable engine modes including a bundled ONNX Runtime wrapper. The new engine supports dynamic shapes, constant folding, and operator fusion, though it is currently CPU-only. Hardware acceleration is redesigned around Universal Intrinsics 2.0, with dedicated HAL support for Intel IPP, Arm KleidiCV (up to 3-4x speedups), Qualcomm FastCV (Hexagon DSP/NPU), and RISC-V Vector. Core improvements include FP16/BF16 data types, true N-dimensional Mat support, up to 2x math performance gains, C++17 as the minimum standard, and NumPy 2.x integration for Python. 3D vision gains multi-camera calibration, point cloud I/O, and dense RGB-D fusion modules. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [OpenCV 5.0 Released With Rewritten DNN Engine, Built-In LLM & VLM Support](https://daily.dev/posts/opencv-5-0-released-with-rewritten-dnn-engine-built-in-llm-vlm-support-dyzhch3c7) · Phoronix · 0 upvotes · 0 comments - [Intel Releases OpenVINO 2026.1 With Backend For Llama.cpp, New Hardware Support](https://daily.dev/posts/intel-releases-openvino-2026-1-with-backend-for-llama-cpp-new-hardware-support-prra96hvb) · Phoronix · 0 upvotes · 0 comments - [oneDNN 3.10 Continues Preparing For Future Intel CPUs With AVX 10.2](https://daily.dev/posts/onednn-3-10-continues-preparing-for-future-intel-cpus-with-avx-10-2-ptkpxduus) · Phoronix · 0 upvotes · 0 comments --- Tags: [#deep-learning](https://daily.dev/tags/deep-learning), [#computer-vision](https://daily.dev/tags/computer-vision), [#opencv](https://daily.dev/tags/opencv) [View this post on daily.dev](https://daily.dev/posts/new-dnn-engine-with-enhanced-onnx-and-llm-vlm-support-intel-arm-and-risc-v-hardware-optimizations-5jrhxdsuq) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"New DNN engine with enhanced ONNX and LLM/VLM support, Intel, Arm, and RISC-V hardware optimizations","url":"https://daily.dev/posts/new-dnn-engine-with-enhanced-onnx-and-llm-vlm-support-intel-arm-and-risc-v-hardware-optimizations-5jrhxdsuq","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/new-dnn-engine-with-enhanced-onnx-and-llm-vlm-support-intel-arm-and-risc-v-hardware-optimizations-5jrhxdsuq"},"datePublished":"2026-06-10T03:32:19.000Z","dateModified":"2026-06-11T00:00:15.064Z","description":"OpenCV 5 has been released with a rebuilt DNN engine that expands ONNX operator coverage from ~22% to over 80%, adds LLM/VLM support (Qwen 2.5, Gemma 3,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/761e2b87bb8a6096ac1c0a895f956f93?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/761e2b87bb8a6096ac1c0a895f956f93?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"CNX Software","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Person","name":"CNXSoft","url":"https://daily.dev/cnxsoft","image":"https://avatars.githubusercontent.com/u/1366138?v=4","interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"EndorseAction"},"userInteractionCount":370}},"commentCount":0,"discussionUrl":"https://daily.dev/posts/new-dnn-engine-with-enhanced-onnx-and-llm-vlm-support-intel-arm-and-risc-v-hardware-optimizations-5jrhxdsuq","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"deep-learning,computer-vision,opencv","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"CNX Software","item":"https://daily.dev/sources/cnx-software"},{"@type":"ListItem","position":3,"name":"New DNN engine with enhanced ONNX and LLM/VLM support, Intel, Arm, and RISC-V hardware optimizations"}]} ```