OpenAI has announced the release of two new open-weight language models—GPT-oss-120b and GPT-oss-20b—marking a significant shift from its closed-source ChatGPT models.
These models can be run locally and customised by users, offering greater flexibility for businesses, developers, and researchers seeking tailored AI solutions.
The larger GPT-oss-120b model is optimised for servers with dedicated graphics cards, while GPT-oss-20b, a more lightweight version, can operate on laptops with just 16GB of RAM. This accessibility makes it easier for individuals and small businesses to implement advanced AI capabilities without relying on cloud-based infrastructure.
Both models support code generation and online research capabilities, similar to existing large language models, but with the added benefit of user-level control over AI parameters. OpenAI has not disclosed the training datasets used for these models.
By launching these models under the "open-weight" category, OpenAI continues to explore new directions in AI transparency and user autonomy, aligning with growing demand for AI tools that can be integrated into private systems and customised for unique workflows.
News Source: Emirates News Agency
