News Summary

OpenAI has launched two new open-weight AI reasoning models, gpt-oss-120b and gpt-oss-20b, available for download on the Hugging Face platform. The models, which are designed to run on different levels of hardware, represent OpenAI’s first open language model release since GPT-2. This move is seen as OpenAI’s response to increasing competition from Chinese AI labs and U.S. governmental encouragement to open source technology. The gpt-oss-120b model is capable of operating on a single Nvidia GPU while the gpt-oss-20b can run on consumer laptops. These models outperform several competitive benchmarks but exhibit higher hallucination rates compared to OpenAI’s proprietary models. The release comes under the Apache 2.0 license, allowing enterprises to monetize the models. Despite the open release, OpenAI withheld the training data, citing ongoing legal issues. The launch aligns with OpenAI’s mission to foster global AGI development based on democratic values.