OpenAI has launched its first open-weight models since 2019: gpt-oss-120b and gpt-oss-20b, offering a shift in AI development strategies. Available under the Apache 2.0 license, these models can be freely downloaded, modified, and deployed without a paid API. The gpt-oss-120b model features 117 billion parameters for high-reasoning tasks, while the 21 billion parameter gpt-oss-20b is designed for consumer-grade hardware. Both models facilitate multi-step reasoning and can be deployed locally or via cloud platforms, enhancing accessibility for developers and enterprises.
Strategic partnerships with Amazon and Microsoft have integrated these models into cloud services, emphasizing collaborative open-weight ecosystems. Although OpenAI filters sensitive training data and has tested the models for misuse, some experts call for further transparency. The models are now accessible for testing on OpenAI’s platform and other services like Hugging Face, marking a pivotal move in bridging proprietary and open-source AI for the industry.
Source link
