GPT-oss is an open-weight, state-of-the-art language model released by OpenAI, available in 20B and 120B parameter versions. It is designed for high performance in reasoning, code generation, and real-world agentic tasks, supporting up to 128k context length and efficient deployment on a range of hardware. The models leverage mixture-of-experts architecture for efficiency and can be fine-tuned for specialized use cases. GPT-oss is suitable for developers and enterprises seeking control, sovereignty, and flexibility in AI deployments, with compatibility for local and cloud environments.