OpenAI Releases Free GPT-OSS Models for Developers to Run Locally
OpenAI launched GPT-OSS, a new set of AI models that developers can download and run for free on their own computers. The company released two versions - a 20 billion parameter model and a 120 billion parameter model - both available through Hugging Face and GitHub.
OpenAI released GPT-OSS, marking the company's first major move into open-weight AI models that anyone can download and use without ongoing costs. The release includes two models: gpt-oss-20b with 20 billion parameters and gpt-oss-120b with 120 billion parameters.
Developers can access the models through Hugging Face, a popular AI model sharing platform, or download them directly from OpenAI's GitHub repository. The models use advanced training techniques focused on reasoning, efficiency, and real-world applications.
To run the models, users need to convert downloaded files into the proper format using OpenAI's provided conversion tools. A detailed coding tutorial shows how to set up the models in Google Colab, a free cloud computing platform, with specific focus on GPU requirements and technical setup.
This release puts OpenAI in direct competition with other open-source AI providers like Meta's Llama models. The move allows developers to choose between OpenAI's paid API services for convenience or these free models for more control and privacy.
This gives small businesses and individual developers access to powerful AI without paying monthly fees to OpenAI. Companies can now run AI models privately on their own servers instead of sending data to OpenAI's cloud services.
Developers will test the models' performance compared to paid alternatives. OpenAI may release additional model sizes based on adoption.
Was this article helpful?
0 people found this helpful