OpenAI Unveils Two Open-Weight AI Reasoning Models to Counter Global Competition On Tuesday, OpenAI announced the release of two open-weight AI reasoning models featuring capabilities comparable to its proprietary o-series. Available for free download on Hugging Face's developer platform, these models have been described by the company as "state-of-the-art" in multiple benchmark tests comparing open models. The new models offer different form factors: the gpt-oss-120b variant capable of running on single Nvidia GPUs, and the lighter gpt-oss-20b model suitable for 16GB consumer laptops. This marks OpenAI's first public language model release since GPT-2 over five years ago. During a technical briefing, OpenAI revealed its open models can offload complex queries to cloud-based AI models, enabling developers to connect open models to proprietary systems for tasks like image processing. While historically favoring closed-source development - which generated significant API revenue - CEO Sam Altman recently acknowledged OpenAI had been "on the wrong side of history" regarding open-source initiatives. The release comes as Chinese AI labs like DeepSeek, Alibaba's Qwen, and Moonshot AI gain momentum in open model development. Government pressures also mount, with the Trump administration encouraging US developers to share more open technologies aligned with democratic values. OpenAI aims to position gpt-oss as a competitive alternative through performance benchmarks showing superior results in coding (Codeforces scores 2622/2516) and general knowledge (19%/17.3% on Human Last Exam) compared to leading open models. However, OpenAI's open models exhibit higher hallucination rates (53%-49% on PersonQA) than their proprietary counterparts. Despite this, the company emphasizes similarities in training methodology with its o-series models, utilizing MoE architecture (activating just 51B parameters per token for 117B gpt-oss-120b) and high-compute reinforcement learning. The models will be released under Apache 2.0 license while keeping training data private due to ongoing copyright litigation. After multiple delayed launches addressing safety concerns, OpenAI found limited risk in potential misuse scenarios. Developers now await DeepSeek R2 and Meta's next open model to see if OpenAI's new offerings maintain their leadership position in open AI space.