openvinotoolkit/openvino

[NPU] Support for qwen3-vl models ExternalIntelPR category: NPU category: NPUW

[NPUW] Support for models with Linear Attention - Part 1 category: Core category: NPU category: NPUW

[GPU][PTL]WA: Cap WG size to 512 for softmax kernels category: GPU

[NPU] Implementing the string-based compiled model compatibility checks category: inference category: CPP API category: NPU

[ITT] Standardize all OV_ITT_DOMAIN display names with ov:: prefix category: inference category: GPU category: CPU category: HETERO category: transformations category: LP transformations ExternalIntelPR category: TEMPLATE category: AUTO category: NPU category: snippets

[DO NOT MERGE] Added condition for adding L0 to compilation when it's actually been used category: build category: dependency_changes do_not_merge

[WIP][Op GroupedMatMul init category: Core category: IE Tests category: docs category: TEMPLATE category: CPP API

Update memory tests no-match-files

[JS] Unskip exportModel()/importModel() tests from basic.test.js category: JS API

Update oneDNN to rls-v3.12 for a custom build category: GPU

Memory optimization during conversion of onnx model OV IR format category: Core category: ONNX FE category: CPP API

Fix NPM vulnerabilities category: JS API

[GPU] Implement by-channel quantized kvcache for Xattetion supported Page Attention category: GPU

[core] Force directory metadata refresh after cache file creation category: inference

Add check for empty constant tensor in expand operation category: ONNX FE

[Draft] Add CPU support for PagedCausalConv1D category: Core category: CPU

[Test] grouped gemm integration category: GPU do_not_review do_not_merge

[TRANSFORMATIONS][GPU] Support 4D InstanceNorm MVN pattern in GroupNormalizationFusion category: transformations ExternalIntelPR

[GPU] Add Prelu and Sign Activation Support for WebNN category: GPU pr: needs tests ExternalIntelPR

[GPU] MoE: supported normalization scale on routing subgraph category: IE Tests category: GPU WIP

[TRANSFORMATIONS] Destroy constants after replacement in ConvertPrecision category: transformations

[NPU] Check the sanity of the device each time the singleton returns the instance created previously no-match-files category: NPU

Cvs 177568 2nd pr category: Core category: IE Tests category: GPU category: CPU category: transformations category: LP transformations

[draft] Fake turboquant category: GPU do_not_review do_not_merge

- Allow quantize fusion for non per-tensor cases if no shape change category: GPU ExternalPR

Fix XAttention BY_TOKEN KV-cache normalization category: GPU

[GPU][PTL] WA: resample_ref max work-group size to 512 to avoid CL_INVALID_WORK_GROUP_SIZE category: GPU

[GPU] onednn integration rls-v3.12 category: GPU

[benchmark_app] Add high precision latency reporting in microseconds category: CPU category: samples ExternalIntelPR

Bump pydata-sphinx-theme from 0.14.4 to 0.17.0 category: docs ExternalPR category: dependency_changes python dependencies