**Rio’s AI Model Falls to Earth, Exposed as a Copycat**
Rio de Janeiro’s IplanRIO team just dropped Rio 3.5 Open 397B, a supposedly state-of-the-art AI model that was supposed to leave all competitors in the dust. Specifically, it was said to top Alibaba’s Qwen 3.7 Plus, which is a major feat in AI research.
But here’s the thing: a company called Nex just came forward and revealed that Rio 3.5 Open 397B was actually built on top of someone else’s work. It’s not like they modified it much – Nex basically said NexMind, their own AI model, was 80% of the code that went into Rio 3.5 Open 397B. That’s a pretty clear case of plagiarism, and it’s not the first time AI researchers have been caught borrowing from others without proper credit.
NexMind is a notable AI model because it’s designed to be highly efficient and was developed with a specific focus on search tasks. **NexMind’s code is available online**, and it seems that Rio’s team just downloaded it and used it as the basis for their supposedly new AI model. Alibaba’s Qwen 3.7 Plus, on the other hand, is a much more complex and original model that was built from scratch.
**What this means**: This whole situation highlights the importance of originality and transparency in AI research. When researchers copy others’ work without permission, it undermines the entire field and makes it harder for people to trust the results. It also means that we won’t learn as much from these new models, because they’re not actually new at all.
