AI industry faces growing divide between open and closed models

The debate between open-source and closed-source artificial intelligence models is evolving as the gap in raw capability narrows. While closed-source models currently dominate top performance leaderboards, open models are seeing high adoption rates among developers.

According to data from a SlashData and Mozilla survey of nearly 1,500 developers, 79% of those adding AI functionality use open models, compared to 71% for closed models. However, a production gap exists: 63% of developers using closed models run them in production, while only 51% of those using open models do the same. Developers using open models frequently cite challenges regarding deployment, scaling, and maintenance rather than model performance.

Safety remains a central point of contention. Companies like Anthropic have raised concerns regarding the risks of releasing open models, whereas other experts argue that transparency in open-source software provides safety advantages. The industry continues to grapple with whether the benefits of accessibility outweigh the potential for misuse by bad actors.