OpenAI has released its first open-weight model, gpt-oss, which is a major release since GPT-2. This new model enables developers and enterprises to run, fine-tune, and deploy OpenAI models under their own conditions, truly achieving controllable AI applications. Users can now run gpt-oss-120b on a single enterprise GPU or run gpt-oss-20b locally, marking further popularization and development of AI technology.
As AI gradually becomes the core of the technology stack, Microsoft is building a comprehensive AI application and intelligent agent factory, aiming to help developers not only use AI but also collaborate with AI for creation. Azure AI Foundry, as a unified platform, provides strong support for building, fine-tuning, and deploying intelligent agents. Meanwhile, Foundry Local brings open-source models to edge devices, making flexible local inference on billions of devices possible.
The release of the two models, gpt-oss-120b and gpt-oss-20b, means users have new choices in terms of performance, convenience, and deployment flexibility. gpt-oss-120b has 120 billion parameters and possesses excellent reasoning capabilities, efficiently handling complex tasks; while gpt-oss-20b is optimized for tool usage and autonomous tasks, suitable for various Windows hardware, especially in environments with limited bandwidth.
Through Azure AI Foundry, developers can easily create inference endpoints, fine-tune models using their own data, and deploy them with high reliability and security. Additionally, Foundry Local supports local inference on personal computers, further enhancing data privacy and control.
This release not only allows developers to gain insight into and customize models but also provides enterprise leaders with flexibility and control. The open nature of gpt-oss means there are no black-box operations, and users can adjust according to their actual needs, ensuring compliance and efficiency of AI applications.
Key Points:
🌟 The release of the gpt-oss model provides developers and enterprises with the ability to run and adjust AI independently.
💻 The gpt-oss-120b and gpt-oss-20b models have excellent performance and are suitable for various application scenarios.
🔒 Azure AI Foundry and Foundry Local offer secure and flexible deployment options, ensuring data privacy.