OpenAI unveils free GPT models for local use
OpenAI has released GPT-OSS, a pair of open-weight language models that can run directly on local machines. Available in 120-billion and 20-billion parameter versions, the models are optimized for deployment on a single Nvidia GPU or even devices with just 16GB of RAM. Both versions are licensed under Apache 2.0, allowing commercial use and full customization. The models are now accessible via Hugging Face, Databricks, AWS, and Azure.
This marks OpenAI’s first open-weight release, signaling a strategic move toward broader accessibility and developer freedom in AI deployment.
About the Author
