Archive — US — techcrunch
OpenAI launches two ‘open’ AI reasoning models
OpenAI has launched two open-weight AI reasoning models, gpt-oss-120b and gpt-oss-20b, available for free on Hugging Face. This marks OpenAI's return to open-sourcing language models after GPT-2, amid rising competition from Chinese AI labs. The models, designed for various hardware capabilities, aim to integrate complex queries with existing AI systems while facing scrutiny over safety and licensing issues.
Key points
- OpenAI launched gpt-oss-120b and gpt-oss-20b, two open-weight AI reasoning models, on Hugging Face, allowing free downloads.
- gpt-oss-120b runs efficiently on a single Nvidia GPU, while gpt-oss-20b is optimized for consumer laptops with 16GB of memory.
- The release is OpenAI's first open-source model since the five-year-old GPT-2, responding to competition from Chinese AI developers.
- The models can communicate with OpenAI's more capable closed models for complex tasks.
- OpenAI aimed to align with the U.S. government's push for open-source AI technology after concerns about Chinese advancements.
- Recent performance tests showed these models hallucinate more than OpenAI’s proprietary models, indicating challenges in accuracy.
- gpt-oss models were trained similarly to proprietary models but won't process images or audio.
Why it matters
- OpenAI's shift back to open source reflects a growing trend in the tech industry toward collaborative innovation and transparency.
- In 2025, open-source AI technologies are expected to flourish, driven by increased competition and the demand for democratic AI solutions.
- The U.S. government's push for open-source AI technology highlights national stability concerns in tech against competitors like China.
OpenAI's launch of open-weight models signifies a pivotal shift towards collaboration in AI development, aiming to reclaim ground amid fierce competition and align with broader technological and political trends by 2025.