Description
OpenAI is releasing gpt-oss-120b and gpt-oss-20b, two state-of-the-art open-weight language models that deliver strong real-world performance at low cost. These models are available under the flexible Apache 2.0 license and outperform similarly sized open models on reasoning tasks. They demonstrate strong tool use capabilities and are optimized for efficient deployment on consumer hardware.
The gpt-oss-120b model achieves near-parity with OpenAI o4-mini on core reasoning benchmarks while running efficiently on a single 80 GB GPU. The gpt-oss-20b model delivers similar results to OpenAI o3-mini on common benchmarks and can run on edge devices with just 16 GB of memory. This makes it ideal for on-device use cases, local inference, or rapid iteration without costly infrastructure. Both models also perform strongly on tool use, few-shot function calling, and chain-of-thought reasoning.
Safety is foundational to OpenAI's approach to releasing models, particularly for open models. The gpt-oss models have undergone comprehensive safety training and evaluations, ensuring they meet high safety standards. OpenAI has also collaborated with early partners to explore real-world applications, providing these models to empower developers, enterprises, and governments to run and customize AI on their own infrastructure.
The gpt-oss models were trained using advanced pre-training and post-training techniques, focusing on reasoning, efficiency, and usability across various deployment environments. Each model is a Transformer that leverages mixture-of-experts to optimize performance. The models support three reasoning efforts—low, medium, and high—allowing developers to adjust performance based on their needs.
The weights for both models are freely available for download on Hugging Face and come natively quantized in MXFP4. This allows for efficient memory usage, making the models flexible and easy to run locally or through third-party providers. OpenAI is committed to fostering a healthy open model ecosystem, encouraging developers and researchers to experiment and innovate with these powerful tools.
gpt-oss Highlights
Open-weight models
Apache 2.0 license
Strong reasoning capabilities
Tool use support
Customizable
Efficient deployment
Supports local inference
Mixture-of-experts architecture
Getting Started with gpt-oss
Access model: Visit the OpenAI website to access gpt-oss.
Authenticate: Set up an account if required for model access.
Set up environment: Prepare your development environment for integration.
Integrate via API: Use the provided API to integrate gpt-oss into your application.
Optimize: Adjust model parameters for performance based on your specific use case.
gpt-oss's Use Cases
- On-device AI
- Local inference
- Rapid iteration
- Tool integration
- Custom AI solutions
