Step-by-Step Guide to Installing OpenAI GPT-OSS AI Model on Windows 11

How to Install the GPT-OSS AI Model on Windows 11

Installing the latest GPT-OSS (Open Source Software) AI model on your Windows 11 machine is a straightforward process. Below, you’ll find detailed instructions to successfully complete this setup.

Getting Started with GPT-OSS

GPT-OSS is a text-based AI model from OpenAI, which comes in two primary versions: gpt-oss-20b (approximately 21 billion parameters) and gpt-oss-120b (around 117 billion parameters).Both models are available under the Apache 2.0 license, facilitating unrestricted downloading, running, modification, and redistribution.

Model Specifications

gpt-oss-20b: This model is lightweight, requiring a minimum of 16GB of memory, making it suitable for standard hardware like laptops and smartphones, and is optimized for low-latency tasks.
gpt-oss-120b: This larger variant excels in complex reasoning tasks and demands high-performance systems, ideally with at least one 80GB GPU for optimal performance.

Both models enhance interactive capabilities, offering features like chain-of-thought reasoning and tool use, along with adaptable reasoning levels (low, medium, high).

What Can Users Do with GPT-OSS?

Tech enthusiasts, developers, and researchers can leverage GPT-OSS to operate a fully functional chatbot on their local machines without a subscription, thus retaining complete control over personal data. The open nature of its Apache 2.0 license enables extensive customization and fine-tuning of the model.

Typical applications include coding assistance, problem-solving, conversational interactions, and facilitating agentic workflows. Moreover, it can function effectively offline, eliminating the need for an internet connection.

Who Benefits from OpenAI GPT-OSS?

The intended audience for these AI models comprises developers, researchers, academics, and technically-inclined individuals. If you’re eager to run these models locally, the Ollama tool serves as an efficient open-source solution compatible with Windows 11 and 10, among other platforms.

While it’s advisable to operate these AI models on systems equipped with a Neural Processing Unit (NPU)—like a Copilot+ PC with a Snapdragon processor—they can be executed on most hardware setups. Notably, Ollama collaborates with Nvidia to enhance GPT-OSS performance on GeForce RTX GPUs.

However, it’s important to note that models may experience significant latency on machines lacking a capable GPU or NPU.

Installation Steps for GPT-OSS on Windows 11

Below are methods to install the GPT-OSS AI model:

Install GPT-OSS on Windows 11 Using Ollama (GUI Method)
Install GPT-OSS on Windows 11 Using Ollama (CLI Method)

Install GPT-OSS Using Ollama (GUI Method)

Follow these steps to install GPT-OSS locally via the GUI:

Open the Start menu on Windows 11.
Search for Command Prompt (or Terminal), right-click the top result, and select Run as administrator.
Execute the following command to install the official Ollama tool:

winget install --id Ollama. Ollama

Click Finish to finalize the installation.
Launch the Ollama app from the Start menu if it does not open automatically.
Select the AI model menu and choose “GPT-OSS:20b”.
Compose a question and submit it to initiate the download of the model.

Upon completion of these steps, you will be ready to utilize the GPT-OSS AI model through your command-line interface.

Install GPT-OSS Using Ollama (CLI Method)

To install GPT-OSS locally using the command line, adhere to these steps:

Open the Start menu.
Search for Command Prompt (or Terminal), right-click the top result, and select Run as administrator.
Input this command to install the Ollama tool:

winget install --id Ollama. Ollama

Finish the installation by clicking Finish.
(Option 1) To install the 20B model locally, enter the following command:

ollama pull gpt-oss:20b

Quick note: This command retrieves the GPT-OSS model with 20 billion parameters. You may also opt for gpt-oss:latest to get the most recent version.

(Option 2) For the larger model, use this command:

ollama pull gpt-oss:120b

(Optional) To verify that the AI model was successfully installed, enter:

ollama list

(Optional) For model details (e.g., configuration and parameters), input:

ollama show gpt-oss:20b

Finally, to run the GPT-OSS model, type:

ollama run gpt-oss:20b

After following these instructions, you will be equipped to utilize the OpenAI GPT AI model directly via the command line.

If you wish to download the Ollama installer, visit the official site or check its GitHub repository. Additionally, the ollama --help command provides insight into available commands, while ollama run --help lists commands specific to a selected model.

Choosing the Right Model

If you’re planning to run these AI models, the gpt-oss-20b variant is recommended due to its reduced hardware demands. Conversely, opting for the gpt-oss-120b requires robust hardware, typically including an NVIDIA GPU boasting at least 80GB VRAM, a high-performance processor (like AMD EPYC or Intel Xeon), a minimum of 256GB RAM, and a high-speed NVMe SSD with at least 500GB of available space for model processing and storage.

Stay Updated with My Newsletter Get the latest guides and updates delivered to your inbox.

For those with an Ollama account, signing in grants access to the built-in web search, enabling the integration of real-time information with the model’s capabilities. The Turbo option is also available for $20 monthly, allowing users to leverage high-performance cloud hardware for running large AI models too hefty for local machines.

The Ollama app currently does not offer an option for pre-downloading AI models, but this can be accomplished through the command-line interface.

For more details and images, visit Source & Images.