# What changed in openvino

> 61 changes in openvino 2026 from its official release notes, upgrade guides and advisories: 8 deprecations, 2 security, 22 new APIs and options. Each one cited. Sources checked 2026-10-18. HTML: https://unstale.sh/pypi/openvino/

To see which of these apply to a repository, run `npx @unstale/cli init` in it.

## openvino 2026

### Security

- Hardened import against malformed or forged blobs with integer-overflow protection, bounded compatibility-string parsing, and forged-blob rejection. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Added index validation to harden the operator against malformed inputs. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)

### Deprecations

- Runtime model reshaping through auto batch size is deprecated and will be removed; dynamic shape models are recommended instead. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- Runtime model reshaping through auto shape is deprecated and will be removed; dynamic shape models are recommended instead. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- Modifying runtime information through read-only ConstOutput objects in the Python API is deprecated; use non-const outputs when updating runtime metadata. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO GenAI deprecates finish_chat(); pass a ChatHistory object directly to generate() instead. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- The legacy graph iteration path enabled with ONNX_ITERATOR=0 is deprecated and will be removed in future releases. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- The dedicated OpenVINO operator for Kubernetes and OpenShift is deprecated in favor of the KServe operator and will no longer be actively developed. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- Support for Python 3.10 is scheduled to end in OpenVINO 2026.5 because Python 3.10 has reached end of life. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO GenAI deprecates start_chat(); pass a ChatHistory object directly to generate() instead. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)

### New APIs and options

- Added partial rotary-embedding support for models using GroupQueryAttention. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- OpenVINO GenAI adds ASRPipeline support in Node.js, providing a JavaScript pipeline API for streaming automatic speech recognition with performance metrics. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- BGE M3 becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- BGE Reranker-V2-M3 becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU and GPU support for DeepSeek OCR-2. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO GenAI previews DFlash acceleration for Qwen models on GPUs. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- Fun-ASR-Nano becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- Granite 4.0 H Tiny becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU and GPU support for Hy-MT2-1.8B. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO extends ITT profiling to NPU execution, allowing Intel VTune Profiler to analyze CPU, GPU, and NPU execution through one toolchain. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU, GPU, and NPU support for Kokoro-82M. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- LFM2.5-8B-A1B becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- MiniCPM5-2B becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO GenAI adds MTP speculative decoding for Gemma 4, Qwen3.5, and Qwen3.6 on CPUs and GPUs to increase throughput and reduce latency without sacrificing accuracy. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU and GPU support for Muse Glimmer 30B. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO Model Server adds support for Muse Glimmer 30B. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- Qwen-Image becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU and GPU support for Qwen3-ASR. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU and GPU support for Qwen3-VL-4B with EAGLE-3. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO Model Server adds support for Qwen3.8 27B. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- RF-DETR becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO GenAI supports Top-K Tree Drafting for EAGLE-3, enabling higher throughput on VLM pipelines than Chain Drafting (Top-1). (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)

### Performance

- OpenVINO Model Server previews idle model management that unloads models when they are not in use to reduce memory usage. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO GenAI optimizes Gemma 4 inference with long-context inputs on Xe3 integrated graphics in Intel Core Ultra Series 3 processors. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)

### Behavior and default changes

- Added causal-attention support for models using GroupQueryAttention. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Preview bounded dynamic-shape support on NPUs extends to vision workloads such as image super-resolution and has been validated with ESPCN. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds NPU support for FLUX.2-Klein 4B. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU support for Gemma-3n. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- ENABLE_CPU_PINNING is now supported on dual-socket Intel® Xeon® 6+ processor Windows Servers. (2026.3.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.3.1)
- Qwen3.8-27B is functionally enabled for use with OpenVINO. (2026.3.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.3.1)

### Fixes and other changes

- Added divide-by-zero and overflow guards to shape inference. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Corrected shape inference for f8e4m3 and other low-precision data types; f8e4m3 Gather is now enabled for the windowed KV-cache. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Fixed incorrect results for sequence models with dynamic shapes and corrected blocked-format reorder offsets on GPU. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Added bfloat16 support and reordered weight, scale, and zero-point layouts to correctly decode more quantized models. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Fixed unexpected crashes on Intel Core Ultra Series 3 Windows systems with certain NPU PV drivers; migration to this release is recommended for affected earlier releases. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Added axes validation to prevent a potential out-of-bounds write. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- The instruction-set cap now works in release builds and is modeled as a feature bitmask. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Aligned JIT kernel selection with the ISA cap so operators produce correct results when the cap is set. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Fixed wrong or corrupted results with remote zero-copy output tensors when a model output has a transpose under dynamic shapes. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Fixed recursive layout and format propagation through Reshape nodes during GPU graph compilation. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- Fixed an out-of-bounds index in the sliding-window KV-cache path that could cause memory corruption or a GPU driver crash. (2026.4.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.1)
- OpenVINO adds CPU and GPU support for Gemma 4 12B. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU and GPU support for Granite 4.0 H Micro. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO adds CPU and GPU support for Qwen3.8 27B. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- OpenVINO GenAI previews visual-token support to reduce latency and speed up GenAI pipelines on Intel Core Ultra Series 3 processors. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- Z-Image-Turbo becomes available as an early release with CPU and GPU support. (2026.4.0) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.4.0)
- The chat template was updated so it can be parsed by the Minja templating engine. (2026.3.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.3.1)
- Fixed an accuracy issue when the input prompt is long. (2026.3.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.3.1)
- Fixed an accuracy issue when an image is included in the input. (2026.3.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.3.1)
- Fixed an accuracy issue for these models when targeting an NPU device. (2026.3.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.3.1)
- Muse-Glimmer-30B is functionally enabled for use with OpenVINO. (2026.3.1) — [release notes](https://github.com/openvinotoolkit/openvino/releases/tag/2026.3.1)