OpenAI has unveiled two open artificial intelligence models, GPT-oss-120b and GPT-oss-20b, that emulate human reasoning. Available on Hugging Face, these models can generate text, perform complex tasks such as coding, and search the internet based on user prompts. Both models are open-weight systems, akin to Meta’s Llama, meaning OpenAI has disclosed their underlying parameters to facilitate customization for developers. However, the company is withholding the training data, which limits their classification as fully open-source AI models. This release comes months after China’s DeepSeek garnered attention with its own AI software, highlighting the evolving landscape of competitive AI development. By making these models accessible and customizable, OpenAI aims to encourage innovation while navigating the challenges of transparency in AI training. Engage with the latest advancements in AI by exploring these innovative models, designed for flexibility and utility in various applications.
Source link
Share
Read more