2026.4.12026-10-01
Summary of major features and improvements
- Improved GroupQueryAttention support for large language models served through the ONNX frontend:
- Added causal-attention and partial rotary-embedding support to GroupQueryAttention, broadening coverage of modern LLMs.
- Extended MatMulNBits with bfloat16 support and reordered weight/scale/zero-point layouts for correct decoding of more quantized models.
- Fixed correctness issues in the sliding-window KV-cache path that could produce wrong results:
- an out-of-bounds ScatterUpdate index that could cause memory corruption or a driver crash on GPUs,
- incorrect Gather shape inference for f8e4m3 and other low-precision data types (f8e4m3 Gather is now enabled for the windowed KV-cache).
- Fixed issues:
- GPU:
- Fixed wrong or corrupted results when using remote (zero-copy) output tensors with a transpose at the model output under dynamic shapes.
- Fixed GRU sequence (RNN) models producing incorrect results with dynamic shapes, and corrected blocked-format reorder offsets.
- Fixed recursive layout/format propagation through Reshape nodes during GPU graph compilation.
- NPU:
- ID 233822: Fixed unexpected application crashes on Intel® Core™ Ultra Series 3 processors on Windows that could occur with certain Intel® NPU PV drivers (32.0.100.4300, 32.0.100.4509, and 32.0.100.4512). It is recommended to migrate to OV 2026.4.1 in case the issue is observed with OV releases 2026.2, 2026.3, or 2026.4.
- Hardened compiled-blob import against malformed or forged blobs: fixed a possible integer overflow in metadata parsing, bounded recursion in the compatibility-string parser, and rejected forged blobs on import.
- CPU:
- The OVCPUMAX_ISA instruction-set cap now works in release builds (previously debug-only) and is modeled as a feature bitmask.
- Aligned JIT kernel selection with the OVCPUMAX_ISA cap so operators produce correct results when the ISA is capped.
- Core / Frontends:
- Hardened operators against malformed inputs: added index validation to ScatterElementsUpdate, divide-by-zero and overflow guards to Col2Im shape inference, and axes validation preventing a potential out-of-bounds write in the ONNX Pad operator.
You can find the OpenVINO™ toolkit 2026.4.1 release here:
- Download archives* with OpenVINO™
- OpenVINO™ for Python: pip install openvino==2026.4.1
Release documentation is available here: <https://docs.openvino.ai/2026>
Release Notes are available here: <https://docs.openvino.ai/2026/about-openvino/release-notes-openvino.html>
*Other names and brands may be claimed as the property of others.
2026.4.02026-09-16
Summary of major features and improvements
* #### More GenAI coverage and framework integrations to minimize code changes
* New models supported:
* On CPU: Gemma-3n
* On CPU, GPU: Kokoro-82M, Qwen3-VL-4B with EAGLE-3, Qwen3-ASR, Muse Glimmer 30B, Qwen3.8 27B, Gemma 4 12B, Hy-MT2-1.8B, DeepSeek OCR-2, and Granite 4.0 H Micro
* On NPUs: FLUX.2-Klein 4B and Kokoro-82M
* Additional CPU and GPU-enabled models available as early releases: Qwen-Image, Z-Image-Turbo, Granite 4.0 H Tiny, Fun-ASR-Nano, LFM2.5-8B-A1B, MiniCPM5-2B, RF-DETR, BGE Reranker-V2-M3, and BGE M3
* #### Broader LLM model support and more model compression techniques
* OpenVINO™ GenAI adds Multi-Token Prediction (MTP) speculative decoding for Gemma 4, Qwen3.5, and Qwen3.6 on CPUs and GPUs, enabling higher throughput and lower latency without sacrificing accuracy.
* Preview: OpenVINO™ GenAI introduces DFlash acceleration for Qwen models on GPUs, and visual-token support to reduce latency and speed up GenAI pipelines on Intel® Core™ Ultra Series 3 processors.
* With Tree Drafting (Top-K) for EAGLE-3 now supported in OpenVINO™ GenAI, developers can unlock higher throughput on VLM pipelines compared to Chain Drafting (Top-1).
* Xe3 integrated graphics optimizations in OpenVINO™ GenAI improve AI inference performance for Gemma 4 models processing long-context inputs on Intel® Core™ Ultra Series 3 processors.
* OpenVINO extends Instrumentation and Tracing Technology (ITT) profiling support to the NPU, enabling developers to use Intel® VTune™ Profiler to analyze CPU, GPU, and NPU execution through one consistent toolchain.
* #### More portability and performance to run AI at the edge, in the cloud, or locally.
* OpenVINO™ GenAI introduces support for ASRPipeline in Node.js, enabling JavaScript developers to run automatic speech recognition (such as Whisper and Qwen3-ASR) with streaming and performance metrics using a pipeline API similar to C++ and Python.
* Preview: Bounded dynamic-shape support on NPUs now extends to vision workloads such as image super-resolution. This capability has been validated with the ESPCN model.
* Preview: OpenVINO™ Model Server now includes preview support for idle model management, which unloads models when they are not in use to reduce memory usage.
* OpenVINO™ Model Server adds support for new agentic models such as Muse Glimmer 30B and Qwen3.8 27B.
Support Change and Deprecation Notices
* Discontinued in 2026.0:
* The deprecated openvino.runtime namespace has been removed. Please use the openvino namespace directly.
* The deprecated openvino.Type.undefined has been removed. Please use openvino.Type.dynamic instead.
* Support for Debian 10 has been discontinued due to the end of its standard support.
* The PostponedConstant constructor signature has been updated for improved usability:
* Old (removed): Callable[[Tensor], None]
* New: Callable[[], Tensor]
* The deprecated OpenVINO™ GenAI predefined generation configs were removed.
* The deprecated OpenVINO GenAI support for whisper stateless decoder model has been removed. Please use a stateful model.
* The deprecated OpenVINO GenAI StreamerBase put method, bool return type for callbacks, and ChunkStreamer class has been removed.
NNCF createcompressed_model() method is now deprecated and removed in 2026. Please use nncf.prune() method for unstructured pruning and nncf.quantize() for INT8 quantization.
* NNCF optimization methods for TensorFlow models and TensorFlow backend in NNCF are deprecated and removed in 2026. It is recommended to use PyTorch analogous models for training-aware optimization methods and OpenVINO™ IR, PyTorch, and ONNX models for post-training optimization methods from NNCF.
* The following experimental NNCF methods are deprecated and removed: NAS, Structural Pruning, AutoML, Knowledge Distillation, Mixed-Precision Quantization, Movement Sparsity.
* CPU plugin now requires support for the AVX2 instruction set as a minimum system requirement. The SSE instruction set will no longer be supported.
* Dropped support for the TensorFlow Serving (TFS) API in OpenVINO Model Server 2026.3. KServe API is recommended for classic model deployments.
* Support for Stateful Models was removed in OpenVINO Model Server 2026.3. These capabilities were originally introduced for Kaldi audio models which are no longer applicable. Current audio model support relies on the OpenAI API and pipelines implemented with the OpenVINO™ GenAI library.
* Support for Python 3.10 will be discontinued in OpenVINO 2026.5, due to its end-of-life (EOL) status.
* Deprecated and to be removed in the future:
* auto shape and auto batch size (reshaping a model in runtime) will be removed in the future. OpenVINO’s dynamic shape models are recommended instead.
Starting with 2026.0 release major internal refactoring of the graph iteration mechanism has been implemented for improved performance and maintainability. The legacy path can be enabled by setting the ONNXITERATOR=0 environment variable. This legacy path is deprecated and will be removed in future releases.
* Modifying runtime information (RTMap) through read-only ConstOutput objects in the Python API is deprecated and will be removed in a 2027.0 release. Users should use non-const outputs when updating runtime metadata; attempts to modify RTMap through ConstOutput now emit a deprecation warning to ease migration.
* OpenVINO Model Server:
* The dedicated OpenVINO operator for Kubernetes and OpenShift is now deprecated in favor of the recommended KServe operator. The OpenVINO operator will remain functional in upcoming OpenVINO™ Model Server releases but will no longer be actively developed. Since KServe provides broader capabilities, no loss of functionality is expected. On the contrary, more functionalities will be accessible and migration between other serving solutions and OpenVINO Model Server will be much easier.
* Directed Acyclic Graph Scheduler will be deprecated in favor of pipelines managed by MediaPipe scheduler and will be removed in 2026.3. That approach gives more flexibility, includes wider range of calculators and has support for using processing accelerators.
* OpenVINO™ GenAI:
startchat() / finish_chat() APIs are deprecated and will be removed in a future major release. Pass a ChatHistory object directly to generate() instead.
Acknowledgements
Thanks for contributions from the OpenVINO developer community:
@ainergiz
@Akihiro-Tabuchi
@aobolensk
@ash1shb1sht
@da3dsoul
@deepak0x
@goyaladitya05
@hrushitfujitsu
@jatinwadhwa921
@Lagmator22
@lxcxjxhx
@mostafafaheem
@NaiveTomcat
@nuclearcat
@odili
@Passavee-Losripat
@samwebster
@vtanisha
You can find OpenVINO™ toolkit 2026.4 release here:
Download archives with OpenVINO™
* OpenVINO™ for Python: pip install openvino==2026.4.0
Release documentation is available here: https://docs.openvino.ai/2026
Release Notes are available here: https://docs.openvino.ai/2026/about-openvino/release-notes-openvino.html
2026.3.12026-08-26
Summary of major features and improvements
* Functionally enabled Muse-Glimmer-30B model:
* Model card and IR available: OpenVINO/Muse-Glimmer-30B-int4-ov · Hugging Face
* Notebooks to try:
* Muse Glimmer-30B multimodal reasoning with OpenVINO™
* Create VLM-powered Chatbot using OpenVINO Generate API
* Blogs:
* Running Meta’s Muse Glimmer 30B Locally with OpenVINO™ GenAI;
* Day 0 Support for Muse Glimmer from Meta on Intel Platforms
* Functionally enabled Qwen3.8-27B model:
* Model card and IR available: OpenVINO/Qwen3.8-27B-int4-ov · Hugging Face
* Notebook to try: Create VLM-powered Chatbot using OpenVINO Generate API
* Blog: Day 0 Support for Qwen3.8–27B: Run the New Model on Intel Agentic PCs with OpenVINO™ from Day One
* Fixed issues:
Property ENABLECPU_PINNING supported now on dual-socket Intel® Xeon® 6+ processors (formerly codenamed Clearwater Forest) Windows Servers.
* Chat template was updated for Gemma 4 models. Now it can be parced by the Minja templating engine.
* Fixed accuracy issue for the Gemma 4 26B A4B model when the input prompt is long.
* Fixed accuracy issue for the Gemma 4 31B model when an image is included in the input.
* Fixed accuracy issue for lfm2-1.2b, lfm2.5-1.2b-instruct models for NPU device target.
You can find OpenVINO™ toolkit 2026.3.1 release here:
Download archives with OpenVINO™
* OpenVINO™ for Python: pip install openvino==2026.3.1
Release documentation is available here: https://docs.openvino.ai/2026
Release Notes are available here: https://docs.openvino.ai/2026/about-openvino/release-notes-openvino.html
2026.3.02026-08-04
Summary of major features and improvements
* #### More GenAI coverage and framework integrations to minimize code changes
* New models supported:
* On CPU, GPU, and NPU: SmolLM3-3B, LFM2-1.2B, LFM2.5-1.2B
* On CPU and GPU: Harrier OSS-v1-0.6B, Qwen3-8B with EAGLE-3, MiniCPM5-1B, FLUX.2-klein-4B
* Extended to GPU and NPU: YOLO26
* Additional models available as early releases on CPU and GPU: Qwen3-ASR, Qwen3-Omni, Gemma-3n, Qwen3-VL-Embedding-8B, Kokoro-82M
* Support for Hugging Face Transformers v5.5, ensuring compatibility with the latest model architectures on Hugging Face
* #### Broader LLM model support and more model compression techniques
* OpenVINO™ GenAI extends the EAGLE-3 speculative decoding pipeline to LLMs and VLMs, enhancing existing continuous batching and adding Top-K sampling to deliver additional token-generation speedups on CPUs, GPUs, and NPUs.
* Lazy weight loading is enabled for IR and ONNX models to automatically select an optimal loading and compilation path, minimizing peak memory usage during model initialization.
* Neural Network Compression Framework (NNCF) now supports FP8 quantization for ONNX models, helping developers realize FP8 performance, accuracy, and memory gains while expanding low-precision inference options for production deployments.
* OpenVINO™ GenAI now supports three additional pipelines: Omni for multimodal workloads, ASR for speech recognition, and Embedding for multimodal embedding generation.
* #### More portability and performance to run AI at the edge, in the cloud, or locally.
* Introducing support for Intel® Xeon® 6+ processors (formerly codenamed Clearwater Forest)
* MoE offloading to disk enabled, allowing 30B MoE models like Qwen3-30B-A3B to run even on devices with 16 GB of memory while maintaining acceptable tokens-per-second (TPS) generation rates.
* OpenVINO™ Model Server simplifies model deployment and unifies REST API endpoints, reducing command complexity while providing standard v1/chat/completions support for easier integration with other serving frameworks.
* OpenVINO™ Model Server further improves stability, performance, and accuracy for LLMs such as Qwen3.5/3.6 with linear attention and extends tool-parser support to MiniCPM5-1B and LFM2.5, improving accuracy and reliability.
Support Change and Deprecation Notices
* Discontinued in 2026.0:
* The deprecated openvino.runtime namespace has been removed. Please use the openvino namespace directly.
* The deprecated openvino.Type.undefined has been removed. Please use openvino.Type.dynamic instead.
* Support for Debian 10 has been discontinued due to the end of its standard support.
* The PostponedConstant constructor signature has been updated for improved usability:
* Old (removed): Callable[[Tensor], None]
* New: Callable[[], Tensor]
* The deprecated OpenVINO™ GenAI predefined generation configs were removed.
* The deprecated OpenVINO GenAI support for whisper stateless decoder model has been removed. Please use a stateful model.
* The deprecated OpenVINO GenAI StreamerBase put method, bool return type for callbacks, and ChunkStreamer class has been removed.
NNCF createcompressed_model() method is now deprecated and removed in 2026. Please use nncf.prune() method for unstructured pruning and nncf.quantize() for INT8 quantization.
* NNCF optimization methods for TensorFlow models and TensorFlow backend in NNCF are deprecated and removed in 2026. It is recommended to use PyTorch analogous models for training-aware optimization methods and OpenVINO™ IR, PyTorch, and ONNX models for post-training optimization methods from NNCF.
* The following experimental NNCF methods are deprecated and removed: NAS, Structural Pruning, AutoML, Knowledge Distillation, Mixed-Precision Quantization, Movement Sparsity.
* CPU plugin now requires support for the AVX2 instruction set as a minimum system requirement. The SSE instruction set will no longer be supported.
* OpenVINO™ migrated builds based on RHEL 8 to RHEL 9.
* Dropped support for the TensorFlow Serving (TFS) API in OpenVINO Model Server 2026.3. KServe API is recommended for classic model deployments.
* Deprecated and to be removed in the future:
* auto shape and auto batch size (reshaping a model in runtime) will be removed in the future. OpenVINO’s dynamic shape models are recommended instead.
Starting with 2026.0 release major internal refactoring of the graph iteration mechanism has been implemented for improved performance and maintainability. The legacy path can be enabled by setting the ONNXITERATOR=0 environment variable. This legacy path is deprecated and will be removed in future releases.
* OpenVINO Model Server:
* The dedicated OpenVINO operator for Kubernetes and OpenShift is now deprecated in favor of the recommended KServe operator. The OpenVINO operator will remain functional in upcoming OpenVINO™ Model Server releases but will no longer be actively developed. Since KServe provides broader capabilities, no loss of functionality is expected. On the contrary, more functionalities will be accessible and migration between other serving solutions and OpenVINO Model Server will be much easier.
* Support for Stateful models will be deprecated. These capabilities were originally introduced for Kaldi audio models which is no longer relevant. Current audio models support relies on the OpenAI API, and pipelines implemented via OpenVINO GenAI library.
* Directed Acyclic Graph Scheduler will be deprecated in favor of pipelines managed by MediaPipe scheduler and will be removed in 2026.3. That approach gives more flexibility, includes wider range of calculators and has support for using processing accelerators.
* OpenVINO™ GenAI:
startchat() / finish_chat() APIs are deprecated and will be removed in a future major release. Pass a ChatHistory object directly to generate() instead.
You can find OpenVINO™ toolkit 2026.3 release here:
Download archives with OpenVINO™
* OpenVINO™ for Python: pip install openvino==2026.3.0
Acknowledgements
Thanks for contributions from the OpenVINO developer community:
@abhijain1204fujitsu
@Akihiro-Tabuchi
@ankitm3k
@aobolensk
@AshutoshSinghIntel
@daehyun99
@ggladilov
@goyaladitya05
@Lagmator22
@pj3iL
@pjordanandrsn
@ssam18
@sshekhar563
@Sujanian1304
@tknose
@vmadananth
@wine99
Release documentation is available here: https://docs.openvino.ai/2026
Release Notes are available here: https://docs.openvino.ai/2026/about-openvino/release-notes-openvino.html
2026.2.12026-06-17
Summary of improvements
* Fixed issue ID 187077: YOLO26 fails to compile on GPU
* Resolved an issue in NPU plugin with shared L0 command queue causing violation of app/inference priorities
You can find OpenVINO™ toolkit 2026.2.1 release here:
Download archives with OpenVINO™
* OpenVINO™ for Python: pip install openvino==2026.2.1
2026.2.02026-05-28
Summary of major features and improvements
* #### More GenAI coverage and framework integrations to minimize code changes
* New models supported: Gemma 4 E2B and Gemma 4 E4B
* Only on CPUs & GPUs: Qwen3-Coder-Next, Qwen3.5, Qwen3.6, Trinity-mini, LFM2-24B-A2B, LFM2-8B-A1B, LFM2.5-350M
* Only on CPUs: YOLO26
* Only on GPUs: Gemma 4 31B and Gemma 4 26B-A4B
* Extended to GPUs: GPT-OSS-120B
* Scaled Dot-Product Attention (SDPA) path support added for LFM2 models
* Support for Hugging Face Transformers v5.0, ensuring compatibility with the latest model architecture for enhanced interoperability.
* #### Broader LLM model support and more model compression techniques
* OpenVINO™ GenAI introduces extension support for loading custom extension libraries and registering unsupported operations via the extensions property. This gives developers the flexibility to run models with custom ops that OpenVINO doesn't support out of the box.
* INT4 KV-cache compression is enabled for GPUs, with substantial memory reduction when KV cache size is significant, such as with large input prompts exceeding 32K tokens.
* OpenVINO GenAI significantly reduces model loading times on GPU when using cache blobs — preventing bottlenecks for multi-stage AI pipelines, including agentic use cases that rely on multiple models.
* Optimized IR read mode with independently managed constant buffers to reduce peak memory usage by avoiding unnecessary duplication of weight data unless required for correctness (Linux support added in this release).
* Preview: Enhanced XAttention accuracy on CPUs and GPUs through by-channel INT8 KV-cache quantization (compared to by-token INT8 KV-cache), matching the default by-channel INT8 KV cache quantization when XAttention is not enabled.
* #### More portability and performance to run AI at the edge, in the cloud, or locally.
* OpenVINO™ GenAI extends its JavaScript API to include a Text-to-Speech pipeline and VLM samples for browser and Node.js developers.
* Prompt Lookup Decoding extended to vision-language pipelines, delivering significantly faster token generation for multimodal workloads on Intel CPUs and GPUs.
* OpenVINO™ GenAI now has a smaller runtime footprint after eliminating ICU DLL dependencies from tokenization, leading to reduced memory usage, faster startup, and easier deployment.
* OpenVINO™ Model Server extends tool-calling support to Qwen 3.5 and 3.6 models to enable agentic AI use cases.
* OpenVINO™ Model Server adds streaming transcription support for speech-to-text, reducing latency for real-time voice applications.
Support Change and Deprecation Notices
* Discontinued in 2026.0:
* The deprecated openvino.runtime namespace has been removed. Please use the openvino namespace directly.
* The deprecated openvino.Type.undefined has been removed. Please use openvino.Type.dynamic instead.
* The PostponedConstant constructor signature has been updated for improved usability:
* Old (removed): Callable[[Tensor], None]
* New: Callable[[], Tensor]
* The deprecated OpenVINO™ GenAI predefined generation configs were removed.
* The deprecated OpenVINO GenAI support for whisper stateless decoder model has been removed. Please use a stateful model.
* The deprecated OpenVINO GenAI StreamerBase put method, bool return type for callbacks, and ChunkStreamer class has been removed.
NNCF createcompressed_model() method is now deprecated and removed in 2026. Please use nncf.prune() method for unstructured pruning and nncf.quantize() for INT8 quantization.
* NNCF optimization methods for TensorFlow models and TensorFlow backend in NNCF are deprecated and removed in 2026. It is recommended to use PyTorch analogous models for training-aware optimization methods and OpenVINO™ IR, PyTorch, and ONNX models for post-training optimization methods from NNCF.
* The following experimental NNCF methods are deprecated and removed: NAS, Structural Pruning, AutoML, Knowledge Distillation, Mixed-Precision Quantization, Movement Sparsity.
* CPU plugin now requires support for the AVX2 instruction set as a minimum system requirement. The SSE instruction set will no longer be supported.
* OpenVINO™ migrated builds based on RHEL 8 to RHEL 9.
manylinux2014 upgraded to manylinux2_28. This aligns with modern toolchain requirements but also means that CentOS 7 will no longer be supported due to glibc incompatibility.
* Deprecated and to be removed in the future:
* Support for Ubuntu 20.04 has been discontinued due to the end of its standard support.
* auto shape and auto batch size (reshaping a model in runtime) will be removed in the future. OpenVINO’s dynamic shape models are recommended instead.
* With the release of Node.js v22, updated Node.js bindings are now available and compatible with the latest LTS version. These bindings do not support CentOS 7, as they rely on newer system libraries unavailable on legacy systems.
Starting with 2026.0 release major internal refactoring of the graph iteration mechanism has been implemented for improved performance and maintainability. The legacy path can be enabled by setting the ONNXITERATOR=0 environment variable. This legacy path is deprecated and will be removed in future releases.
* OpenVINO Model Server:
* The dedicated OpenVINO operator for Kubernetes and OpenShift is now deprecated in favor of the recommended KServe operator. The OpenVINO operator will remain functional in upcoming OpenVINO Model Server releases but will no longer be actively developed. Since KServe provides broader capabilities, no loss of functionality is expected. On the contrary, more functionalities will be accessible and migration between other serving solutions and OpenVINO Model Server will be much easier.
* TensorFlow Serving (TFS) API support is planned for deprecation. With increasing adoption of the KServe API for classic models and the OpenAI API for generative workloads, usage of the TFS API has significantly declined. Dropping date is to be determined based on the feedback, with a tentative target of mid-2026.
* Support for Stateful models will be deprecated. These capabilities were originally introduced for Kaldi audio models which is no longer relevant. Current audio models support relies on the OpenAI API, and pipelines implemented via OpenVINO GenAI library.
* Directed Acyclic Graph Scheduler will be deprecated in favor of pipelines managed by MediaPipe scheduler and will be removed in 2026.3. That approach gives more flexibility, includes wider range of calculators and has support for using processing accelerators.
* OpenVINO™ GenAI:
startchat() / finish_chat() APIs are deprecated and will be removed in a future major release. Pass a ChatHistory object directly to generate() instead.
You can find OpenVINO™ toolkit 2026.2 release here:
Download archives with OpenVINO™
* OpenVINO™ for Python: pip install openvino==2026.2.0
Acknowledgements
Thanks for contributions from the OpenVINO developer community:
@abbyssoul
@aobolensk
@AshutoshSinghIntel
@dbermond
@desertfury
@FarseenSh
@goyaladitya05
@KarSri7694
@Lagmator22
@mostafafaheem
@om4rrr
@pjordanandrsn
@reeseliao
@Sahilbhatane
@ssam18
@sshekhar563
@Vishwa2684
@vmadananth
Release documentation is available here: https://docs.openvino.ai/2026
Release Notes are available here: https://docs.openvino.ai/2026/about-openvino/release-notes-openvino.html
2026.1.02026-04-07
Summary of major features and improvements
* #### More GenAI coverage and framework integrations to minimize code changes
* New models supported on CPUs & GPUs: Qwen3 VL
* New models supported on CPUs: GPT-OSS 120B
* Preview: Introducing the OpenVINO backend for llama.cpp, which enables optimized inference on Intel CPUs, GPUs, and NPUs. Validated on GGUF models such as Llama-3.2-1B-Instruct-GGUF, Phi-3-mini-4k-instruct-gguf, Qwen2.5-1.5B-Instruct-GGUF, and Mistral-7B-Instruct-v0.3.
* New notebook: Unified VLM chatbot with video file support and interactive model switching across Qwen3-VL, Qwen2.5-VL, and LLaVa-NeXT-Video.
* #### Broader LLM model support and more model compression techniques
* OpenVINO™ GenAI adds TaylorSeer Lite caching for image and video generation, accelerating diffusion-transformer inference across Flux, SD3, and LTX-Video pipelines, aligned with Hugging Face Diffusers.
* LTX-Video generation on GPU achieves end-to-end acceleration through fusion of RMSNorm and RoPE operators, significantly improving video generation performance.
* OpenVINO™ GenAI adds dynamic LoRA support for Qwen3-VL and VL models with LLM, allowing developers to swap adapters at runtime for efficient serving of multiple model variants in production without reloading the base model.
* Preview: The release-weights API for ov::Model enables memory reclamation during model compilation on NPUs, delivering dramatically lower peak memory consumption for edge and client deployments. Users must set this property in ov::Model, and it will be applied during compilation.
* #### More portability and performance to run AI at the edge, in the cloud, or locally.
* Introducing support for Intel® Core™ Series 3 processors (formerly codenamed Wildcat Lake) and Intel® Arc™ Pro B70 Graphics with 32GB memory for single-GPU inference on 20-30B parameter LLMs
* Prompt Lookup Decoding extended to vision-language pipelines, delivering significantly faster token generation for multimodal workloads on Intel CPUs and GPUs.
* OpenVINO™ GenAI now has a smaller runtime footprint after eliminating ICU DLL dependencies from tokenization, leading to reduced memory usage, faster startup, and easier deployment.
* OpenVINO GenAI introduces WhisperPipeline for Node.js via its NPM package, delivering production-ready speech recognition with word-level audio-to-text transcription.
* OpenVINO™ Model Server enhances support for Qwen3-MOE and GPT-OSS-20b models, delivering improved performance, accuracy, and robust concurrent request handling with continuous batching. These pre-optimized models are available on Hugging Face for easy deployment. Additionally, the Model Server introduces image inpainting and outpainting capabilities via the /image endpoint for AI image editing.
Support Change and Deprecation Notices
* Discontinued in 2026.0:
* The deprecated openvino.runtime namespace has been removed. Please use the openvino namespace directly.
* The deprecated openvino.Type.undefined has been removed. Please use openvino.Type.dynamic instead.
* The PostponedConstant constructor signature has been updated for improved usability:
* Old (removed): Callable[[Tensor], None]
* New: Callable[[], Tensor]
* The deprecated OpenVINO™ GenAI predefined generation configs were removed.
* The deprecated OpenVINO GenAI support for whisper stateless decoder model has been removed. Please use a stateful model.
* The deprecated OpenVINO GenAI StreamerBase put method, bool return type for callbacks, and ChunkStreamer class has been removed.
NNCF createcompressed_model() method is now deprecated and removed in 2026. Please use nncf.prune() method for unstructured pruning and nncf.quantize() for INT8 quantization.
* NNCF optimization methods for TensorFlow models and TensorFlow backend in NNCF are deprecated and removed in 2026. It is recommended to use PyTorch analogous models for training-aware optimization methods and OpenVINO™ IR, PyTorch, and ONNX models for post-training optimization methods from NNCF.
* The following experimental NNCF methods are deprecated and removed: NAS, Structural Pruning, AutoML, Knowledge Distillation, Mixed-Precision Quantization, Movement Sparsity.
* CPU plugin now requires support for the AVX2 instruction set as a minimum system requirement. The SSE instruction set will no longer be supported.
* OpenVINO™ migrated builds based on RHEL 8 to RHEL 9.
manylinux2014 upgraded to manylinux2_28. This aligns with modern toolchain requirements but also means that CentOS 7 will no longer be supported due to glibc incompatibility.
* MacOS x86 is no longer supported.
* APT & YUM Repositories Restructure: Starting with release 2025.1, users can switch to the new repository structure for APT and YUM, which no longer uses year-based subdirectories (like “2025”). The old (legacy) structure is unavailable starting 2026.0. Detailed instructions are available on the relevant documentation pages:
* Installation guide - yum
* Installation guide - apt
* OpenCV binaries removed from Docker images.
* Deprecated and to be removed in the future:
* Support for Ubuntu 20.04 has been discontinued due to the end of its standard support.
* auto shape and auto batch size (reshaping a model in runtime) will be removed in the future. OpenVINO’s dynamic shape models are recommended instead.
* With the release of Node.js v22, updated Node.js bindings are now available and compatible with the latest LTS version. These bindings do not support CentOS 7, as they rely on newer system libraries unavailable on legacy systems.
Starting with 2026.0 release major internal refactoring of the graph iteration mechanism has been implemented for improved performance and maintainability. The legacy path can be enabled by setting the ONNXITERATOR=0 environment variable. This legacy path is deprecated and will be removed in future releases.
* OpenVINO Model Server:
* The dedicated OpenVINO operator for Kubernetes and OpenShift is now deprecated in favor of the recommended KServe operator. The OpenVINO operator will remain functional in upcoming OpenVINO Model Server releases but will no longer be actively developed. Since KServe provides broader capabilities, no loss of functionality is expected. On the contrary, more functionalities will be accessible and migration between other serving solutions and OpenVINO Model Server will be much easier.
* TensorFlow Serving (TFS) API support is planned for deprecation. With increasing adoption of the KServe API for classic models and the OpenAI API for generative workloads, usage of the TFS API has significantly declined. Dropping date is to be determined based on the feedback, with a tentative target of mid-2026.
* Support for Stateful models will be deprecated. These capabilities were originally introduced for Kaldi audio models which is no longer relevant. Current audio models support relies on the OpenAI API, and pipelines implemented via OpenVINO GenAI library.
* Directed Acyclic Graph Scheduler will be deprecated in favor of pipelines managed by MediaPipe scheduler and will be removed in 2026.3. That approach gives more flexibility, includes wider range of calculators and has support for using processing accelerators.
* OpenVINO™ GenAI:
startchat() / finish_chat() APIs are deprecated and will be removed in a future major release. Pass a ChatHistory object directly to generate() instead.
You can find OpenVINO™ toolkit 2026.1 release here:
* [Do
2026.0.02026-02-23
Summary of major features and improvements
* #### More GenAI coverage and framework integrations to minimize code changes
New models supported on CPUs & GPUs: GPT-OSS-20B, MiniCPM-V-45-8B, and MiniCPM-o-2.6
* New models supported on NPUs: MiniCPM-o-2.6. In addition, NPU support is now available on Qwen2.5-1.5B-Instruct, Qwen3-Embedding-0.6B, Qwen-2.5-coder-0.5B.
* OpenVINO™ GenAI now adds word-level timestamp functionality to the Whisper Pipeline on CPUs, GPUs, and NPUs, enabling more accurate transcriptions and subtitling in line with OpenAI and FasterWhisper implementations.
* Phi-3-mini FastDraft model is now available on Hugging Face to accelerate LLM inference on NPUs. FastDraft optimizes speculative decoding for LLMs.
* #### Broader LLM model support and more model compression techniques
* With the new int4 data-aware weight compression for 3D MatMuls, the Neural Network Compression Framework enables MoE LLMs to run with reduced memory, bandwidth, and improved accuracy compared to data-free schemes-delivering faster, more efficient deployment on resource-constrained devices.
* Preview: the Neural Network Compression Framework now supports per-layer and per-group Look-Up Tables (LUT) for FP8-4BLUT quantization. This enables fine-grained, codebook-based compression that reduces model size and bandwidth while improving inference speed and accuracy for LLMs and transformer workloads.
* #### More portability and performance to run AI at the edge, in the cloud, or locally.
* Preview: OpenVINO™ GenAI adds VLM pipeline support to enhance Agentic AI framework integration.
* OpenVINO GenAI now supports speculative decoding for NPUs, delivering improved performance and efficient text generation through a small draft model that is periodically validated by the full-size model.
* Preview: NPU compiler integration with the NPU plugin enables ahead-of-time and on-device compilation without relying on OEM driver updates. Developers can enable this feature for a single, ready-to-ship package that reduces integration friction and accelerates time-to-value.
* OpenVINO™ Model Server adds enhanced support for audio endpoint plus agentic continuous batching and concurrent runs for improved LLM performance in agentic workflows on Intel CPUs and GPUs.
Support Change and Deprecation Notices
* Discontinued in 2026.0:
* The deprecated openvino.runtime namespace has been removed. Please use the openvino namespace directly.
* The deprecated openvino.Type.undefined has been removed. Please use openvino.Type.dynamic instead.
* The PostponedConstant constructor signature has been updated for improved usability:
* Old (removed): Callable[[Tensor], None]
* New: Callable[[], Tensor]
* The deprecated OpenVINO™ GenAI predefined generation configs were removed.
* The deprecated OpenVINO GenAI support for whisper stateless decoder model has been removed. Please use a stateful model.
* The deprecated OpenVINO GenAI StreamerBase put method, bool return type for callbacks, and ChunkStreamer class has been removed.
NNCF createcompressed_model() method is now deprecated and removed in 2026. Please use nncf.prune() method for unstructured pruning and nncf.quantize() for INT8 quantization.
* NNCF optimization methods for TensorFlow models and TensorFlow backend in NNCF are deprecated and removed in 2026. It is recommended to use PyTorch analogous models for training-aware optimization methods and OpenVINO™ IR, PyTorch, and ONNX models for post-training optimization methods from NNCF.
* The following experimental NNCF methods are deprecated and removed: NAS, Structural Pruning, AutoML, Knowledge Distillation, Mixed-Precision Quantization, Movement Sparsity.
* CPU plugin now requires support for the AVX2 instruction set as a minimum system requirement. The SSE instruction set will no longer be supported.
* OpenVINO™ migrated builds based on RHEL 8 to RHEL 9.
manylinux2014 upgraded to manylinux2_28. This aligns with modern toolchain requirements but also means that CentOS 7 will no longer be supported due to glibc incompatibility.
* MacOS x86 is no longer supported.
* APT & YUM Repositories Restructure: Starting with release 2025.1, users can switch to the new repository structure for APT and YUM, which no longer uses year-based subdirectories (like “2025”). The old (legacy) structure is unavailable starting 2026.0. Detailed instructions are available on the relevant documentation pages:
* Installation guide - yum
* Installation guide - apt
* OpenCV binaries removed from Docker images.
* Deprecated and to be removed in the future:
* Support for Ubuntu 20.04 has been discontinued due to the end of its standard support.
* auto shape and auto batch size (reshaping a model in runtime) will be removed in the future. OpenVINO’s dynamic shape models are recommended instead.
* With the release of Node.js v22, updated Node.js bindings are now available and compatible with the latest LTS version. These bindings do not support CentOS 7, as they rely on newer system libraries unavailable on legacy systems.
Starting with 2026.0 release major internal refactoring of the graph iteration mechanism has been implemented for improved performance and maintainability. The legacy path can be enabled by setting the ONNXITERATOR=0 environment variable. This legacy path is deprecated and will be removed in future releases.
* OpenVINO Model Server:
* The dedicated OpenVINO operator for Kubernetes and OpenShift is now deprecated in favor of the recommended KServe operator. The OpenVINO operator will remain functional in upcoming OpenVINO Model Server releases but will no longer be actively developed. Since KServe provides broader capabilities, no loss of functionality is expected. On the contrary, more functionalities will be accessible and migration between other serving solutions and OpenVINO Model Server will be much easier.
* TensorFlow Serving (TFS) API support is planned for deprecation. With increasing adoption of the KServe API for classic models and the OpenAI API for generative workloads, usage of the TFS API has significantly declined. Dropping date is to be determined based on the feedback, with a tentative target of mid-2026.
* Support for Stateful models will be deprecated. These capabilities were originally introduced for Kaldi audio models which is no longer relevant. Current audio models support relies on the OpenAI API, and pipelines implemented via OpenVINO GenAI library.
* Directed Acyclic Graph Scheduler will be deprecated in favor of pipelines managed by MediaPipe scheduler and will be removed in 2026.3. That approach gives more flexibility, includes wider range of calculators and has support for using processing accelerators.
You can find OpenVINO™ toolkit 2026.0 release here:
Download archives with OpenVINO™
* OpenVINO™ for Python: pip install openvino==2026.0.0
Acknowledgements
Thanks for contributions from the OpenVINO developer community:
@Sujanian1304
@Nishant-ZFYII
@srinjoydutta03
@samthakur587
@IamRam3
Release documentation is available here: https://docs.openvino.ai/2026
Release Notes are available here: https://docs.openvino.ai/2026/about-openvino/release-notes-openvino.html
2026.0.12026-02-23
* Preview: NPU compiler integration with the NPU plugin enables ahead-of-time and on-device compilation without relying on OEM driver updates. This feature is enabled by default in this release package.
* Known Issues
* Component: optimum; ID: 179936
Description: phi-4-multimodal instruct model isn’t functional when converted using optimum-cli as channel-wise one (with –group-size -1) with OpenVINO 2026.0. It’s recommended to use for the conversion OV 2025.4/OV 2025.4.1
* Component: GenAI ; ID: 179754
Description: Text2VideoPipeline will crash with RuntimeError when calling generate(guidancescale=1.0) after the model was reshaped or compiled with guidancescale > 1.0; use guidance_scale >= 1.0001 as a workaround until the fix is available.
* Component: Runtime; ID: 180693
Description: Qwen3-30B-A3B converted with newer transformers doesn’t work, recommend using transformers 4.55.4 for model conversion which was verified and worked.
* Component: GenAI ; ID: 179973
Description: Qwen2-vl, Qwen-2.5VL, Qwen3-VL dense models may not work through GenAI API with GPU, due internal issue on model transformation level
* Component: Runtime ; ID: 180696
Description: 2nd (and further) latency degradation for Qwen3-MOE family, including lack of ability to fit a model on iGPU, due high memory consumption and potential graph corruption. Problem affects only IRs generated with 2026.0, former IRs generated with 2025.4 will work properly.
* Component: Runtime ; ID: 179009
Description: Memory leak for static builds with HybridCRT enabled; impacts Windows only
You can find OpenVINO™ toolkit 2026.0.1 release here:
Download archives with OpenVINO™
2025.4.12025-12-18
> NOTE: Please continue using OpenVINO 2025.4 release unless you require the specific bug fixes addressed in the 2025.4.1 version.
Summary of improvements
* Preview: Mixture of Experts (MoE) models optimized for CPUs and GPUs, validated for GPT-OSS 20B model.
How to convert model: optimum-cli export openvino -m "openai/gpt-oss-20b" out_dir --weight-format int4
* Fixed issue ID 174531: Accuracy regression of Mistral-7b-instruct-v0.2 and Mistral-7b-instruct-v0.3 on all devices when executed with OpenVINO GenAI. As a workaround, use the IR converted with OpenVINO 2025.3.
* Fixed issue ID 176777: Using the callback parameter with the Python API call generate() in Text2ImagePipeline, Image2ImagePipeline, InpaintingPipeline may cause the process to hang. As a workaround, do not use the callback parameter. C++ implementations was not affected.
Resolved an issue in the NPU plugin where the Level Zero (L0) context was implemented as a static global object and only destroyed during DLL unload, even after unloadplugin() was called. This behavior prevented the driver from spawning threads required for certain optimizations and features.
You can find OpenVINO™ toolkit 2025.4.1 release here:
Download archives with OpenVINO™
* OpenVINO™ for Python: pip install openvino==2025.4.1