Citation and evidence

OpenVINO

22 min full readUpdated 33 references

This article's verification

Report a problem with this article

Edit article
More

Use this article

Raw MarkdownExplore connections

Improve this page

Revision historyDiscuss this article

Topics

AI InferenceDeveloper ToolsOpen Source AI

Cite this article

OpenVINO (Open Visual Inference and Neural Network Optimization) is an open-source toolkit developed by Intel for optimizing and deploying deep learning inference across a heterogeneous set of Intel hardware targets, including x86 and Arm CPUs, integrated and discrete GPUs, and Neural Processing Units (NPUs).[1] First released on May 16, 2018 as a computer-vision toolkit for edge devices, OpenVINO has expanded into a general-purpose inference runtime that supports models trained in PyTorch, TensorFlow, ONNX and other frameworks, and is now Intel's principal software vehicle for generative AI on client and edge silicon.[2][3] The official repository describes it succinctly as an "Open-source software toolkit for optimizing and deploying deep learning models," with documented use across computer vision, automatic speech recognition, generative AI, and natural language processing.[1] The toolkit is distributed under the Apache License 2.0 and developed in the open at github.com/openvinotoolkit/openvino.[4] Beyond the core runtime, the OpenVINO ecosystem includes the Neural Network Compression Framework (NNCF) for quantization and pruning, an OpenVINO Tokenizers extension, the OpenVINO Model Server (OVMS), and an OpenVINO GenAI library targeted at LLM and vision-language workloads.[5][6][7][8]

OpenVINO logo
OpenVINO logo. Intel Corporation SVG file: Valohai development team, via Wikimedia Commons, Apache License 2.0.

Infobox

FieldValue
Original releaseMay 16, 2018 (Intel Distribution of OpenVINO Toolkit)
DeveloperIntel Corporation
LicenseApache License 2.0
Source repositorygithub.com/openvinotoolkit/openvino
Latest release (as of November 2026)2026.4.1 (October 1, 2026)
LanguagesC++ core, Python and C bindings
Supported devicesIntel CPU (x86, Arm), Intel GPU (integrated, Arc discrete), Intel NPU (Core Ultra)
Model intakeOpenVINO IR, ONNX, PyTorch, TensorFlow, TensorFlow Lite, PaddlePaddle, JAX
Hugging Face integrationoptimum-intel

Expanded article table

What does OpenVINO stand for?

The acronym OpenVINO expands to Open Visual Inference and Neural Network Optimization, reflecting the toolkit's original 2018 framing as a visual, computer-vision product. Despite the "Visual" in the name, OpenVINO has since grown well beyond vision into general LLM, speech and generative-AI inference.

When was OpenVINO released?

Origins and the 2018 launch

The Intel Distribution of OpenVINO toolkit was first announced and released on May 16, 2018, when Intel said it was "designed to fast-track development of high-performance computer vision and deep learning inference applications at the edge," targeting vision-centric workloads in cameras, IoT devices, and embedded systems.[2] The initial release framed OpenVINO as the successor to the older Intel Computer Vision SDK, bundling a Model Optimizer that consumed models from popular open-source frameworks such as Caffe, TensorFlow and MXNet, an Inference Engine, and pre-trained models from the Open Model Zoo. It targeted Intel CPUs and integrated GPUs along with specialized accelerators including FPGA cards and Movidius vision processing units (VPUs).[2][9]

The Movidius VPU branch entered the lineage through Intel's September 2016 acquisition of the Irish chip designer Movidius, co-founded in 2005 in Dublin by Sean Mitchell, David Moloney and Val Muresan. Movidius produced the Myriad 2 and Myriad X VPUs, low-power chips for computer vision that Intel later folded into OpenVINO as the MYRIAD plugin and offered as Neural Compute Stick devices. The Myriad 2-based Fathom USB stick ran at 80 to 150 GFLOPS on about 1 W of power, and the third-generation Myriad X added a dedicated Neural Compute Engine for deep-learning inference.[9] The same architectural team's "Keem Bay" / Vision Processing Unit work, detailed publicly in 2019, eventually evolved into the Neural Processing Unit (NPU) shipped in Intel Core Ultra client processors.[10]

2019 to 2022: stabilization and LTS cadence

Across 2019, 2020 and 2021 Intel iterated the toolkit on a roughly quarterly cadence, with releases such as 2019 R1, 2020.1, 2021.1 and 2021.4 progressively broadening framework coverage and Linux distribution support.[11] The core code had been open-sourced early: the repository now at openvinotoolkit/openvino on GitHub dates from October 2018 and is licensed under the Apache License 2.0, with the binary Intel Distribution continuing alongside it.[1][4]

The 2022.1 release, shipped in March 2022, was a substantial rewrite that introduced the IRv11 intermediate representation and dropped support for the older IRv10 in the Post-Training Optimization Tool (POT); it expanded natural-language model coverage with more than 70 additional INT8-quantized models, moved Ubuntu 18.04 to "supported with limitations" status, and unified the device-portability model so that the same compiled graph could move more cleanly between CPU, GPU and accelerator targets.[12] The 2022.x line also added or improved frontends for TensorFlow Lite and PaddlePaddle, broadening the "intake" side of the conversion pipeline beyond the traditional TensorFlow, Caffe, MXNet and ONNX lineup of the 2018 release.[11]

In late 2022, Intel designated 2022.3 LTS as a Long-Term Support release with extended maintenance, VPU support, missing from the initial 2022.3.0 release, was restored in the 2022.3.1 LTS update (June 2023) and kept in 2022.3.2 (May 2024), the last releases to support the Neural Compute Stick 2 / Movidius Myriad X. The 2023.x and later release lines do not support Movidius VPU accelerators.[11][13]

2023 to 2026: PyTorch frontend, NPU, and generative AI

OpenVINO 2023.0 introduced direct conversion from PyTorch model objects into OpenVINO Intermediate Representation without the intermediate step of ONNX export, an option that has since become the recommended path for PyTorch users.[14] In the same period, the Model Optimizer command-line tool was officially deprecated in favor of the new OpenVINO Model Converter (OVC) Python API; Model Optimizer and the separate openvino-dev package were removed with the 2025.0 release.[15]

The 2024.0 release shipped on March 6, 2024 and added a preview NPU plugin for the integrated Neural Processor Unit in Intel Core Ultra processors (codename Meteor Lake), distributed through the main openvino PyPI package; the same release tightened the integration with the Hugging Face ecosystem, including INT4 weight quantization for popular LLMs.[3] Subsequent 2024.x and 2025.x releases extended NPU coverage to the Core Ultra Series 2 ("Lunar Lake") generation, which features a more capable six-engine NPU (compared with two compute engines on the original Meteor Lake NPU) and introduced NormalFloat 4-bit (NF4) weight quantization as the first data type supported only on the newer NPU.[16] OpenVINO GenAI, the dedicated generative-AI library, has had its own repository since October 2023; its 2026.0 release (February 2026) added a Text-to-Video generation pipeline for the LTX-Video model, and the library had reached version 2026.4.1.0 by October 2026.[7][21]

A condensed timeline of headline releases is summarized below:

ReleaseYearNotable change
2018 (initial)2018Original public release; CPU, integrated GPU, FPGA, Movidius VPU
2019 R12019Refreshed Model Optimizer; expanded Open Model Zoo
2021.1 / 2021.42021Broader Linux distribution coverage
2022.12022IRv11; rewritten Inference Engine API; +70 INT8 models
2022.3 LTS2022Long-term support designation
2022.3.1 / 2022.3.2 LTS2023-2024Last releases supporting Movidius VPU / NCS2
2023.02023Direct PyTorch frontend; Movidius VPU support removed
2024.02024NPU plugin in main PyPI package; Hugging Face INT4 quantization
2025.02025Model Optimizer and openvino-dev package removed; OVC API only
2026.0 / 2026.12026Enhanced GenAI; preview Text2Video via LTX-Video; OVMS 2026.1

Expanded article table

How does OpenVINO work?

Architecture

OpenVINO is organized around a core C++ runtime with Python and C bindings, plus pluggable per-device backends. A typical workflow has three stages: convert a trained model from a source framework (PyTorch, TensorFlow, ONNX, PaddlePaddle, JAX) into the OpenVINO format, compile it for a target device using core.compile_model, and then run inference through the resulting CompiledModel and inference requests.[14][17] The same Python or C++ application can dispatch the same compiled graph to different physical devices by changing a device string such as CPU, GPU, or NPU, and the runtime supports heterogeneous execution that splits a graph across multiple devices.[17]

What is the OpenVINO Intermediate Representation (IR)?

OpenVINO's native graph format is the Intermediate Representation (IR), a pair of files comprising an XML description of the graph topology and a binary .bin file holding weights. IR is produced either by the Model Optimizer (in 2022 and earlier) or by the OpenVINO Model Converter API (openvino.convert_model / ovc from 2023 onward).[14][15] IRv11, introduced in 2022.1, is the current major IR version; legacy IRv10 graphs are no longer accepted by NNCF or the POT toolchain.[12]

In addition to its own IR, the OpenVINO runtime can directly load and execute ONNX graphs through an ONNX frontend without an intermediate IR conversion, and from 2023.0 onward it can directly ingest a PyTorch nn.Module via the PyTorch frontend, internally building an OpenVINO graph through tracing rather than first exporting to ONNX. The PyTorch path is now considered the primary conversion route for PyTorch models, and converting via ONNX is documented as a fallback when direct conversion fails.[14]

What hardware does OpenVINO support?

OpenVINO exposes hardware through a plugin architecture; the principal first-party plugins are:

Plugin / device stringHardware
CPUIntel x86 CPUs (AVX2, AVX-512, AMX) and Arm CPUs (NEON, SVE)
GPUIntel integrated GPUs and discrete GPUs, including the Intel Arc family
NPUIntel NPU integrated into Core Ultra (Meteor Lake), Core Ultra Series 2 (including Lunar Lake) and, from release 2025.4, Core Ultra Series 3 processors
MYRIAD (legacy)Movidius Myriad VPUs and Neural Compute Stick 2 (through 2022.3.x LTS only)
FPGA (legacy)Intel Arria/Stratix FPGAs via bitstreams shipped through the 2018 "with FPGA Support" packages

Expanded article table

The Movidius VPU plugin was retired with the cancellation of VPU accelerator support after the 2022.3.x LTS line.[11][13] FPGA support, present in the 2018 "OpenVINO Toolkit for Linux with FPGA Support" SKUs that shipped bitstreams for Arria 10 cards, was discontinued in subsequent releases as Intel reorganized its programmable-logic strategy.[18]

The NPU plugin shipped initially as a preview in 2023.3 and became part of the main pip install openvino distribution starting with 2024.0; on Core Ultra Series 2 (Lunar Lake) the same plugin exposes the larger six-engine NPU and supports additional low-precision data types including NF4 weight quantization that are not available on the original Meteor Lake NPU.[3][16]

How does OpenVINO optimize models? Quantization and the NNCF

Model compression in OpenVINO is delegated to the Neural Network Compression Framework, an open-source Python package hosted at github.com/openvinotoolkit/nncf and licensed under Apache 2.0.[5] NNCF was introduced in an Intel-authored arXiv paper, "Neural Network Compression Framework for fast model inference," submitted on February 20, 2020 by Alexander Kozlov, Ivan Lazarevich, Vasily Shamporov, Nikolay Lyalyushkin and Yury Gorbachev, and last revised in December of that year.[19] The paper describes NNCF as a framework for compressing neural networks with optional fine-tuning, implementing sparsity, quantization and binarization methods designed to produce models that run efficiently on general-purpose hardware (CPU, GPU) and on specialized deep-learning accelerators while preserving the original accuracy.[19]

The framework implements both post-training algorithms (8-bit Post-Training Quantization, weights compression for LLMs, activation sparsity) and training-time algorithms (Quantization-Aware Training, weight-only quantization with LoRA and Neural Low-rank Search, structured pruning), and it accepts models in PyTorch, TorchFX, ONNX and OpenVINO formats with OpenVINO as the preferred backend for post-training quantization.[5] NNCF's PTQ workflow requires only a model and a small calibration dataset on the order of 300 samples to apply 8-bit quantization.[5]

For LLM weight compression specifically, NNCF supports data-free INT8 and INT4 weight quantization as well as data-aware methods such as AWQ (Activation-aware Weight Quantization), GPTQ, quantization scale estimation, and mixed-precision quantization; on Core Ultra Series 2 NPUs it additionally supports the NormalFloat 4-bit (NF4) data type.[24][16] These algorithms are exposed both directly through NNCF and indirectly through the optimum-intel Hugging Face integration, which uses NNCF as its compression backend for OpenVINO IR and ONNX models.[6] The repository's examples include post-training quantization with accuracy control for a YOLOv8 object detector.[5]

The PyPI package is published as nncf. As of November 2026 the latest release is v3.4.0, dated September 17, 2026.[5]

What are OpenVINO Tokenizers?

OpenVINO Tokenizers is a separately maintained extension at github.com/openvinotoolkit/openvino_tokenizers that adds text-processing operators to the runtime so that tokenization and detokenization can be expressed as OpenVINO graphs without external dependencies. The extension converts Hugging Face tokenizers into OpenVINO models, supports BPE, WordPiece, SentencePiece, Tiktoken and RWKV tokenizer types, and can be fused with downstream LLM graphs to produce a single end-to-end pipeline. Once exported, both the tokenizer and detokenizer become first-class OpenVINO models that can be read, compiled and saved with the standard runtime APIs, and they can also be combined with the model itself into a single graph; the documentation calls out classifiers and RAG embedders as cases where this fused form is preferred because both tokenizer and model are invoked once per inference. The extension additionally provides a "greedy decoding" pipeline component for simple text-generation use cases.[20]

The package serves as the tokenizer backbone for OpenVINO GenAI: its openvino_tokenizers check command, which compares a converted tokenizer against the original Hugging Face tokenizer, runs all five of its validation steps automatically when openvino_genai is installed. As of November 2026 the latest release is 2026.4.1.0, dated October 1, 2026, and the project is licensed under Apache 2.0.[20]

What is OpenVINO GenAI?

OpenVINO GenAI is a higher-level library, distributed from github.com/openvinotoolkit/openvino.genai, that runs on top of the OpenVINO Runtime and implements generative-AI pipelines for LLMs, vision-language models, image generation, speech recognition and speech synthesis. It exposes both C++ and Python APIs and implements supporting infrastructure such as continuous batching, prefix caching and KV-cache management. The library's stated motivation is that it offers "better performance than Python-based runtimes" by minimizing the per-token overhead in the inference loop, while preserving a Python-friendly facade for rapid prototyping.[7][32]

Concrete supported workloads include LLM chat (Llama, Phi, Mistral and Qwen families), VLM image and video analysis (LLaVA, MiniCPM-V, Qwen2-VL, Phi-3.5-Vision, InternVL2), image generation with Stable Diffusion and Flux, Whisper-based speech recognition, SpeechT5 text-to-speech, semantic-search embedding generation and text reranking for retrieval workflows.[7] OpenVINO 2026.1 also introduced, as a preview, an OpenVINO backend for llama.cpp enabling optimized inference on Intel CPUs, GPUs and NPUs, validated on GGUF builds of models such as Llama-3.2-1B-Instruct and Mistral-7B-Instruct-v0.3, and added a notebook for a unified VLM chatbot with video-file support and interactive model switching across Qwen3-VL, Qwen2.5-VL and LLaVa-NeXT-Video.[11]

GenAI also implements the Speculative Decoding pipeline as a serving-side optimization, exposing draft-and-target model pairs through OVMS for higher effective throughput on LLM workloads where draft acceptance rates are favorable.[23]

What is the OpenVINO Model Server (OVMS)?

The OpenVINO Model Server is a standalone serving system written in C++ that hosts OpenVINO-optimized models behind network APIs, hosted at github.com/openvinotoolkit/model_server under Apache 2.0.[8] It is API-compatible with TensorFlow Serving's gRPC interface and with the KServe v2 (REST and gRPC) inference protocol, exposing a unified Predict and ModelMetadata surface that lets existing clients of those ecosystems run inference on OpenVINO-optimized models without code changes.[8][22] Beyond classical predict APIs, OVMS exposes an OpenAI-compatible HTTP API for LLM serving and integrates the continuous-batching pipeline from OpenVINO GenAI to serve LLM and VLM workloads.[23] OVMS supports loading models stored locally, on object storage, or pulled from the Hugging Face Hub, and is distributed as Docker images for both bare-metal and Kubernetes deployments. As of November 2026 the latest release is v2026.4.1, dated October 6, 2026.[8]

The server's performance-tuning documentation advises REST users to send binary data where possible, to reduce communication and JSON deserialization overhead.[33]

How does OpenVINO integrate with Hugging Face (optimum-intel)?

optimum-intel is a Hugging Face-maintained Python package that connects the Transformers, Diffusers, Sentence Transformers and timm libraries with OpenVINO; earlier versions also integrated the Intel Neural Compressor and Intel Extension for PyTorch.[25] The Intel and Hugging Face collaboration was announced in mid-2022, and the initial OpenVINO integration into optimum-intel landed on November 2, 2022, based on OpenVINO 2022.2.[6] The package introduces OVModelForXxx wrapper classes that mirror Transformers' AutoModelForXxx classes but execute inference through OpenVINO Runtime, and it provides convenience APIs for exporting Hugging Face Hub models to OpenVINO IR and for invoking NNCF-backed quantization.[25] In an early benchmark cited by Hugging Face, applying NNCF quantization through optimum-intel to a Vision Transformer reduced memory footprint by 3.8x (344MB to 90MB) and improved per-sample latency by 2.4x (98ms to 41ms), with negligible accuracy change (the quantized and original models both reached 87.6% accuracy on the same task).[6]

The integration was deliberately staged: the November 2022 launch focused on encoder-style models (BERT, DistilBERT and their kin) where post-training static quantization and quantization-aware training are well understood, and subsequent releases broadened to encoder-decoder, decoder-only and diffusion architectures. More recent integrations cover weight-only INT4 and NF4 quantization of LLMs and pipelines that combine optimum-intel export with OpenVINO GenAI inference for end-to-end deployment, including the option to push the resulting compressed model back to the Hugging Face Hub for sharing.[6][24]

Hugging Face's documentation presents optimum-intel as a near drop-in path to Intel acceleration: users replace from transformers import AutoModelForXxx with from optimum.intel import OVModelForXxx, change the model class name, and the rest of the standard Transformers and pipelines API continues to work, with OpenVINO performing inference under the hood on Intel CPU, GPU or NPU targets.[25]

How fast is OpenVINO? Benchmarks and performance

OpenVINO publishes a regularly updated set of platform benchmarks through the OpenVINO Model Hub, covering throughput and latency for representative computer-vision and generative workloads across Intel CPUs, GPUs and NPUs. The internal benchmark_app tool, distributed with the runtime, lets users reproduce these measurements on local hardware.[26]

In standardized industry benchmarks, Intel reported on May 5, 2025 that its Core Ultra Series 2 processors were the first to achieve full Neural Processing Unit support in the MLPerf Client v0.6 benchmark from MLCommons (see MLPerf); the submission used the Llama 2 7B model and recorded a first-token latency of 1.09 seconds and an NPU throughput of 18.55 tokens per second. Intel attributed the result to joint optimization between its NPU hardware teams and the OpenVINO software stack.[27] Jon Peddie Research's coverage of the results added that the integrated Intel Arc GPU posted a faster time-to-first-token than other vendors' results.[28]

For Lunar Lake (Core Ultra Series 2), OpenVINO documentation notes that systems may need more than 16 GB of RAM to process prompts longer than 1024 tokens with 7B-class models such as Llama-2-7B, Mistral-0.2-7B and Qwen-2-7B, and requires LLMs exported for the NPU to use symmetric 4-bit (INT4 or NF4) weight compression.[16]

What is OpenVINO used for? Applications

OpenVINO is widely deployed as an inference backend for Edge AI applications because of its support for Intel-class CPUs and integrated GPUs that are common in industrial PCs, retail kiosks, security cameras and consumer laptops. Documented uses include medical imaging at the edge: for example, an Intel video published by the Edge AI and Vision Alliance walks through kidney CT scan analysis in the GE Healthcare Clinical Care Suite, an AI healthcare solution powered by OpenVINO.[29] On the consumer side, QNAP offers the OpenVINO Workflow Consolidation Tool, an App Center application that turns its network-attached storage appliances into inference servers for OpenVINO-accelerated vision models such as image classification Convolutional Neural Networks.[30]

In the generative-AI era, OpenVINO has been positioned as the principal runtime for "AI PC" workloads on Intel client silicon, integrating with the llama.cpp ecosystem and exposing both LLM and VLM pipelines through OpenVINO GenAI; the 2026.0 release added a preview Text2Video pipeline using the LTX-Video model.[21] On the server side, OVMS is used to serve OpenVINO-optimized models in Kubernetes and OpenShift clusters; Intel's documentation describes it as suited to microservice deployments in Kubernetes and OpenShift clusters, with horizontal and vertical inference scaling.[22]

How is OpenVINO used in academic research?

NNCF, which is part of the OpenVINO toolkit ecosystem, is itself the subject of an Intel-authored research paper (arXiv:2002.08679), establishing a citable primary source for downstream academic work that uses OpenVINO-compatible compression methods.[19]

What are the limitations of OpenVINO?

OpenVINO has historically been most performant on Intel hardware, and while CPU support extends to Arm processors, performance optimization on non-Intel platforms is less mature than on Intel-specific paths. The retirement of Movidius VPU support after the 2022.3.x LTS line removed a class of dedicated low-power vision accelerators that some industrial customers had built systems around, leaving those users to either freeze on the LTS branch or migrate to NPU-equipped Core Ultra processors.[13] FPGA support was likewise discontinued after the 2018 generation of "with FPGA Support" packages.[18]

On the NPU side, real-world reports note that the integrated NPU in some Core Ultra configurations can be slower than the same machine's CPU for certain LLM workloads, with users filing performance issues against the openvinotoolkit/openvino repository documenting these cases; appropriate device selection between CPU, GPU and NPU therefore remains workload-dependent.[31] The end-of-life of the standalone openvino-dev package and the Model Optimizer CLI in the 2025 release line is a backwards-incompatible change that has required users to migrate scripts to the new openvino.convert_model API.[15]

How does OpenVINO compare to TensorRT and ONNX Runtime?

ToolkitPrimary vendorNative graph formatHardware focusLicense
OpenVINOIntelOpenVINO IR (XML + bin)Intel CPU, GPU, NPUApache 2.0
TensorRTNVIDIATensorRT engineNVIDIA GPU (CUDA)Proprietary
ONNX RuntimeMicrosoft and communityONNXCPU, GPU, NPU via execution providersMIT
TensorFlow Lite (LiteRT)GoogleTFLite FlatBufferMobile and embedded (Arm CPU, NPU, GPU)Apache 2.0
llama.cppOpen-source communityGGUFCPU, plus GPU/NPU backendsMIT

Expanded article table

Compared with NVIDIA's TensorRT, OpenVINO targets a broader set of CPU-class devices and is open source, but it does not run on NVIDIA GPUs. Compared with ONNX Runtime, OpenVINO can act as an ONNX Runtime execution provider but also exposes its own IR and runtime APIs that allow more aggressive Intel-specific optimizations.[14] On the serving side, OVMS competes with TensorFlow Serving, the KServe stack, BentoML and Ray Serve for the "production model server" role; its differentiator is that the bundled runtime is OpenVINO and it is API-compatible with both TensorFlow Serving and KServe.[22]

Continue reading

More from See also

Topics:AI InferenceDeveloper ToolsOpen Source AI

References

  1. ^1 ^2 ^3openvinotoolkit, "openvino: OpenVINO is an open source toolkit for optimizing and deploying AI inference", GitHub, 2026-04-07. github.com/...openvino. Accessed 2026-05-21.
  2. ^1 ^2 ^3Khari Johnson, "Intel launches OpenVINO computer vision toolkit for edge computing", VentureBeat, 2018-05-16. venturebeat.com/...sion-toolkit-for-edge-computing. Accessed 2026-05-21.
  3. ^1 ^2 ^3Intel Corporation, "Release Notes for Intel Distribution of OpenVINO Toolkit 2024.0", Intel Developer Zone, 2024-03-06. intel.com/...2024-0. Accessed 2026-05-21.
  4. ^1 ^2openvinotoolkit, "openvino/LICENSE at master", GitHub, 2024. github.com/...LICENSE. Accessed 2026-05-21.
  5. ^1 ^2 ^3 ^4 ^5 ^6openvinotoolkit, "nncf: Neural Network Compression Framework for enhanced OpenVINO inference", GitHub, 2026-04-08. github.com/...nncf. Accessed 2026-05-21.
  6. ^1 ^2 ^3 ^4 ^5Hugging Face, "Accelerate your models with Optimum Intel and OpenVINO", Hugging Face Blog, 2022-11-02. huggingface.co/...openvino. Accessed 2026-05-21.
  7. ^1 ^2 ^3 ^4openvinotoolkit, "openvino.genai: Run Generative AI models with simple C++/Python API and using OpenVINO Runtime", GitHub, 2026-04-07. github.com/...openvino.genai. Accessed 2026-05-21.
  8. ^1 ^2 ^3 ^4openvinotoolkit, "model_server: A scalable inference server for models optimized with OpenVINO", GitHub, 2026-04-07. github.com/...model_server. Accessed 2026-05-21.
  9. ^1 ^2Wikipedia contributors, "Movidius", Wikipedia, 2024. en.wikipedia.org/...Movidius. Accessed 2026-05-21.
  10. ^Kyle Wiggers, "Intel details Keem Bay, a next-gen Movidius Myriad VPU for computer vision at the edge", VentureBeat, 2019-11-12. web.archive.org/...for-computer-vision-at-the-edge. Accessed 2026-11-05.
  11. ^1 ^2 ^3 ^4 ^5openvinotoolkit, "Releases openvinotoolkit/openvino", GitHub, 2026-04-07. github.com/...releases. Accessed 2026-05-21.
  12. ^1 ^2Intel Corporation, "Release Notes for Intel Distribution of OpenVINO Toolkit 2022", Intel Developer Zone, 2022. intel.com/...2022-2. Accessed 2026-05-21.
  13. ^1 ^2 ^3Intel Community, "NEW LTS RELEASE: Intel Distribution of OpenVINO toolkit 2022.3.1 LTS", Intel Community, 2023. community.intel.com/...1499215. Accessed 2026-05-21.
  14. ^1 ^2 ^3 ^4 ^5OpenVINO documentation, "Convert a PyTorch Model to OpenVINO IR", docs.openvino.ai, 2024. docs.openvino.ai/...pytorch-to-openvino-with-output. Accessed 2026-05-21.
  15. ^1 ^2 ^3Intel Corporation, "Release Notes for Intel Distribution of OpenVINO Toolkit 2025.0", Intel Developer Zone, 2025. intel.com/...2025-0. Accessed 2026-05-21.
  16. ^1 ^2 ^3 ^4OpenVINO documentation, "OpenVINO GenAI on NPU", docs.openvino.ai, 2025. docs.openvino.ai/...inference-with-genai-on-npu. Accessed 2026-05-21.
  17. ^1 ^2OpenVINO documentation, "NPU Device", docs.openvino.ai, 2024. docs.openvino.ai/...npu-device. Accessed 2026-05-21.
  18. ^1 ^2intel-iot-devkit, "openvino-with-fpga-hello-world-face-detection", GitHub, 2018. github.com/...with-fpga-hello-world-face-detection. Accessed 2026-05-21.
  19. ^1 ^2 ^3Alexander Kozlov, Ivan Lazarevich, Vasily Shamporov, Nikolay Lyalyushkin, Yury Gorbachev, "Neural Network Compression Framework for fast model inference", arXiv:2002.08679, 2020-02-20. arxiv.org/...2002.08679. Accessed 2026-05-21.
  20. ^1 ^2openvinotoolkit, "openvino_tokenizers: OpenVINO Tokenizers extension", GitHub, 2026-04-07. github.com/...openvino_tokenizers. Accessed 2026-05-21.
  21. ^1 ^2OpenVINO toolkit, "OpenVINO 2026.0: New Models, Enhanced GenAI, and Smarter Compression", Medium / OpenVINO toolkit blog, 2026. medium.com/...and-smarter-compression-bf846a59cda8. Accessed 2026-05-21.
  22. ^1 ^2 ^3OpenVINO documentation, "OpenVINO Model Server: What is OpenVINO Model Server", docs.openvino.ai, 2025. docs.openvino.ai/..._what_is_openvino_model_server. Accessed 2026-05-21.
  23. ^1 ^2OpenVINO documentation, "Efficient LLM Serving", docs.openvino.ai, 2025. docs.openvino.ai/...ovms_docs_llm_reference. Accessed 2026-05-21.
  24. ^1 ^2Hugging Face, "Optimize and deploy with Optimum-Intel and OpenVINO GenAI", Hugging Face Blog, 2024. huggingface.co/...deploy-with-openvino. Accessed 2026-05-21.
  25. ^1 ^2 ^3huggingface, "optimum-intel: Optimum Intel: Accelerate inference with Intel optimization tools", GitHub, 2024. github.com/...optimum-intel. Accessed 2026-05-21.
  26. ^OpenVINO toolkit, "Introducing OpenVINO Model Hub: Benchmark AI Inference with Ease", Medium / OpenVINO toolkit blog, 2024. medium.com/...-ai-inference-with-ease-2cd7ad8f5e4d. Accessed 2026-05-21.
  27. ^Intel Newsroom, "Intel Achieves First, Only Full NPU Support in MLPerf Client v0.6 Benchmark", Intel, 2025-05-05. newsroom.intel.com/...mlperf-client-v0-6-benchmark. Accessed 2026-05-21.
  28. ^Jon Peddie Research, "Intel the first company to attain NPU support in MLPerf Client v0.6 benchmark", JPR, 2025. jonpeddie.com/...t-in-mlperf-client-v0-6-benchmark. Accessed 2026-05-21.
  29. ^Edge AI and Vision Alliance, "Real-time Medical Imaging Using OpenVINO", Edge AI and Vision Alliance, 2022-04. edge-ai-vision.com/...dical-imaging-using-openvino. Accessed 2026-05-21.
  30. ^QNAP Systems, "OpenVINO on NAS: Accelerate and verify deep learning inference", QNAP, 2024. qnap.com/...openvino. Accessed 2026-05-21.
  31. ^openvinotoolkit, "[Performance]: 265k NPU much slower than CPU at LLM inference (Issue #32306)", GitHub, 2025. github.com/...32306. Accessed 2026-05-21.
  32. ^OpenVINO documentation, "Generative AI workflow", docs.openvino.ai, 2025. docs.openvino.ai/...openvino-workflow-generative. Accessed 2026-11-05.
  33. ^OpenVINO documentation, "Performance tuning" (OpenVINO Model Server), docs.openvino.ai, 2025. docs.openvino.ai/...ovms_docs_performance_tuning. Accessed 2026-11-05.

Improve this article

Add missing citations, update stale details, or write a clearer explanation. Anyone can edit: changes go live when you publish, and every edit is checked by an editor.

7 revisions by 1 contributor · v7 · 4,315 words · full history

Fact-checks are independent of edits: a reviewer re-verifies the article against its sources and stamps the date. How we verify

Research and drafting on this wiki are AI-assisted, under named human editorial standards. How AI is used here

Reviewer note: Independent fact-check of the full article against its cited sources (150 claims, 36 sources fetched); 8 material + 25 minor fixed

Cite this page: AI Wiki. "OpenVINO." aiwiki.ai, updated 28 Oct 2026, fact-checked 4 Nov 2026. CC BY 4.0. https://aiwiki.ai/wiki/openvino

Edit article