What's the difference between closed, open‑source and open-weight AI? A researcher explains

Understanding the distinctions between closed, open-source, and open-weight artificial intelligence models is crucial as AI technology rapidly advances. These labels refer not to the AI's personality, but to the accessibility and modifiability of its underlying information and code.

The concept of open-source software, which emphasizes the right of users to run, study, modify, and distribute software, originated in the free software movement of the 1980s and '90s. Key principles included making the source code publicly available. Over time, various open-source licenses, such as the GNU General Public License and the MIT License, were developed to govern usage and distribution.

In the realm of artificial intelligence, particularly large language models like ChatGPT, the open-source debate has resurfaced. Meta's release of its LLaMa model in February 2023, providing the inference source code and learned weights, was a significant step. However, organizations like the Open Source Initiative have noted that LLaMa's licensing may restrict commercial use, questioning its 'true' open-source status.

In response, companies are releasing "open weight" models, such as DeepSeek AI's DeepSeek and Alibaba's Qwen. These models offer less restrictive terms for reuse, leading to rapid adoption within the AI community. This distinction between open-source and open-weight is becoming increasingly important as developers navigate the landscape of AI accessibility and proprietary control.