openvinotoolkit/openvino
- Stars
- 11,548
- Language
- C++
- License
- Apache-2.0
- Daily
- rank 8 · +246 stars gained
Repository metrics were observed on this edition date and may have changed.
Key Details OpenVINO is a toolkit for converting models from multiple training frameworks and deploying them to inference environments. Converted models can be deployed without the original training framework, and it provides inference paths for CPUs and supported Intel GPUs and NPUs. Use Cases The README's Python example converts a PyTorch or TensorFlow model into an OpenVINO model, compiles it for a CPU target, and runs inference. It also points to examples for chatbots, object detection, and Whisper speech recognition. Costs and Limitations Apache-2.0 code is provided, but operating costs cannot be determined from the license alone. Local and self-hosted hardware and cloud resource costs are separate, and model-specific terms of use and pricing terms for additional services need to be checked. Adoption Considerations Check support requirements for model operations, precision, devices, and drivers. The README's list of supported devices does not guarantee that a particular model will run or achieve a given level of performance. Performance and usage telemetry collection and opt-out settings also need review. Why It Stands Out It was observed at No. 8 on daily Trending, with a gain of 246 stars. Version 2026.4.1, listed on 2026-10-01, includes expanded ONNX GroupQueryAttention support, quantized model handling, and fixes for incorrect GPU results, NPU crashes, and input validation issues.
View on GitHub