OpenAI unveils its first publicly available artificial intelligence in six years: guide to downloading it

The world of Machine Learning has just experienced a revolution. After six years of waiting, OpenAI has finally unveiled its very first artificial intelligence open to the general public. Imagine the impact of having open-weighted language models at your fingertips, accessible to everyone. Whether you’re a researcher, developer, or entrepreneur, this announcement could transform the way you interact with artificial intelligence. Ready to explore this new territory? Here’s a simple guide to downloading these powerful models.

This Tuesday, OpenAI launched its new open-weighted AI models, a significant advancement that promises to revolutionize the accessibility of artificial intelligence. The gpt-oss-120b and gpt-oss-20b models, designed for bothinnovationand efficiency, are now available to everyone, from researchers to startups, without the need for advanced expertise. Let’s discover together how to download and take advantage of these new resources. New AI Models: What Does It Mean?OpenAI has made a bold choice by opting for

open-weight language models, a first since the release of GPT-2 in 2019. With this initiative, the company is making parameters available that allow developers to explore, test, and adapt them for their specific needs. No access to the original training data is required, which greatly simplifies the use of these models.

Differences with Open Source It’s essential to understand thatopen-weight does not mean open source. While open-source models allow anyone to freely study, modify, and share them, OpenAI limits its offering to publishing only the weights.

These numerical values are crucial for running the model, but the source code and training data are not publicly available. This represents a cautious approach, but it still opens the door to many possibilities for innovation.

Download the models easily For those who want to experiment with these new models , downloading is very simple. Head to Hugging Face where you can access gpt-oss-120b and gpt-oss-20b in just a few clicks. These models are already quantized in MXFP4 format, ensuring optimal performance while consuming fewer resources. The best part? It’s completely free.Compatibility and technical recommendations Keep in mind that each model has its own requirements. Thegpt-oss-120B

requires more powerful hardware configurations, recommending at least a graphics card with 80 GB of memory. In contrast, the

gpt-oss-20B is much more accessible and can run on a standard PC with at least 16 GB of RAM. This flexibility allows a wide range of users to benefit from artificial intelligence, even without cutting-edge hardware.Optimal Use of Models

These AI models can be deployed not only on your

personal computer but also via online artificial intelligence solutions. OpenAI has collaborated with giants like Hugging Face, Azure, AWS, and Databricks to facilitate their integration into various environments. If you want to work offline, gpt-oss-20B is designed to operate without an internet connection. Security Measures in Place OpenAI takes security

very seriously. Before their launch, these models underwent rigorous testing, including a version intentionally tweaked for malicious purposes. This proactive approach is integral to the development of these technologies, ensuring that users can explore these models with confidence.

For those of you interested in further exploring the implications of these advances, I recommend checking out some relevant articles like Llama-4or how to leverage detectable AI

To read OpenClaw et Claude Code : transformez votre assistant IA en podcasteur personnel sur Spotify

here

.