OpenAI’s new open source language model GPT -OOSS‑20BCan now be operated locally by Windows 11 users through Microsoft’s Windows AI Foundry platform. The model requiring a computer with 16 GB Vram offers a solution that can be used in offline for developers.
The model is optimized to be effective in tasks such as text creation and vehicle integration. Especially in developer -oriented applications, results can be produced without being connected to the network connection thanks to local executive capabilities. Microsoft has made this model integrated with vehicles such as Visual Studio Code and Windows Terminal. The installation process can be completed with a few commands via Winget. With this step, a model of OpenAI becomes available locally on Windows devices.
GPT -OOSS‑20BIt is one of the first models published by OpenAI as open-weighted. The model, which is shared with the Apache 2.0 license, allows both individual and corporate developers to re -adjust them. It is stated that the model has given low -delayed results in areas such as code writing, mathematical calculation and health data analysis. However, it is also included in the official documents that the model can produce 53 %incorrect information in questions containing person data. Therefore, it is recommended to verify the outputs of the model in sensitive content.
Microsoft has made GPT -Ooss –20b local running with Foundry Local
The Windows AI Foundry platform is located as the desktop version of Azure AI Foundry and is designed to be used in non -online systems. This platform supports model training, update, version tracking and model export. GPT -OOSS‑20B In addition to the model, a larger version GPT -OOSS‑120B It was announced with the same open license. Microsoft says that MacOS support will be added in the future. It has also been confirmed that AMD’s new generation systems with RYZEN AI supported this model can run with high performance.
GPT -OOSS‑20Bwhile giving similar results with the O3 – Minini model, it can appeal to different uses thanks to its interactive work with tools used in software development processes. It cannot interact with image, audio and video data, but it can do task -oriented transactions with multiple command chains. Thanks to the installation guidelines offered by Microsoft, users can run the model over the command line. This means that the model can be used not only on the cloud, but also in local networks and independent systems.
Amazon is among the platforms supported by this model. Accessible via Amazon Bedrock GPT -OOSS Models are presented outside the OpenAI and Microsoft’s private cooperation frame. This development is a significant change in terms of making OpenAI open to more platforms. The fact that both Microsoft and Amazon quickly integrate these models into their systems show that manufacturer artificial intelligence models can work with wider access.
All weights of the model can be downloaded for free via Huging Face. In addition, detailed comparison tables and performance comparisons on the OpenAI’s official website have been published. All these developments allow the developer community to use modern language models in local systems. Microsoft’s steps in this field will continue to open more local AI models on the Windows platform.