The open-source model Rio 3.5 397B, launched by an IT company under the government of Rio de Janeiro, Brazil, claimed to surpass Qwen 3.7 Plus in multiple benchmark tests, achieving SOTA and ranking among the global top tier. However, within less than 24 hours, Nex-AGI, an AI agent open-source project alliance initiated by Shanghai Chuangzhi Institute, issued a statement accusing Rio 3.5 of being a patchwork model. The Nex team conducted mathematical analysis and comparison of Rio 3.5's weights, revealing its underlying technological formula as: Rio 3.5 ≈ 0.6 × Nex N2 Pro + 0.4 × Qwen 3.5. This means that 60% of its core genetic makeup comes from the Nex N2 Pro previously open-sourced by the Nex team, with the remaining 40% derived from Alibaba's Qwen 3.5. The Nex team provided two independent verification methods: First, after removing Rio's hardcoded system prompt 'You are Rio,' the deployed model had a 79% probability of identifying itself as 'Nex from Nex-AGI' and a 0% probability of claiming to be Rio. It even recited verbatim the background story specifically customized by Nex-AGI. Second, every weight tensor in Rio exhibited the same 0.6/0.4 mixture ratio across all 60 layers and every component of the network, with statistical deviations reaching thousands of standard deviations. Other fine-tuned models could not be explained using this interpolation method. The Nex team stated that Rio 3.5's use of Nex N2 Pro to piece together SOTA performance precisely demonstrated the strength of Nex's open-source foundation. They also emphasized that the open-source community operates under its own rules, welcoming everyone to use it, but attribution and acknowledgment are uncrossable bottom lines.