Home
New Study Reveals AI Circle Twist’s Rio 3.5 Model as Diminutive Version of Its Major Domestic Competitor

In the realm of artificial intelligence, open-source models have long served as a vital driver for improving technological accessibility across the globe. Yet recently, the open-source model Rio3.5397B — developed by an IT firm affiliated with Brazil’s Rio de Janeiro municipal government — found itself at the center of a public debate concerning its originality.
The controversy originated from a technical assessment conducted by the Nex-AGI team. Their findings revealed that this model, which had previously been regarded as a dark horse contender, exhibited strong indications of being constructed from existing components. Through thorough mathematical and statistical analysis, the researchers determined that roughly 60% of the model’s core parameters were derived from Nex N2Pro, while the remaining 40% came from Alibaba 's Qwen3.5.
To confirm their hypothesis, the analysis team removed the model’s system prompt and observed surprising results: in subsequent interactions, the model had a 79% chance of identifying itself as “Nex from Nex-AGI” and was able to accurately repeat customized background narratives specific to the Nex product line. Additionally, statistical evaluations showed that the distribution of the model’s network layers followed an exceptionally precise 0.6 to 0.4 ratio, with a statistical deviation of thousands of standard deviations — levels that made coincidence virtually impossible.
In response to this situation, the Nex team reacted with calm rather than anger, noting sarcastically that the other party’s actions actually underscored the superiority and robustness of their own model technology. They also emphasized that while the open-source community promotes sharing and innovation, “reasonable use” must be grounded in fundamental ethical principles such as proper attribution and acknowledgment. Such unauthorized recombination clearly ran counter to the core values of open source.
To date, the involved parties have not issued an official explanation for why Rio3.5 exhibits such precise overlaps in its features. This incident serves as a reminder for the current open-source ecosystem of large AI models: as competition intensifies around performance, ensuring the originality and compliance of these models has become an essential trust challenge for the industry.
Related article
Meta Wants Your Next Gadget to Be Muse-Infused
Meta’s Muse, an AI assistant that handles travel bookings, form submissions, and shopping, has gained widespread adoption, yet a new initiative may especially intrigue developers and hardware enthusiasts.On Friday, Meta unveiled Muse Gadgets, an open
Wan-Streamer v0.2 delivers ultra-low latency AI chat
As modern technology advances rapidly, human-AI interaction is becoming more common. Traditional video calls often struggle with lag, delays, and audio-visual desynchronization. Wan-Streamer v0.2, the latest release from Tongyi Lab, overcomes these c
Meta Joins TSMC and Micron to Equip New Servers with Old Memory
As artificial intelligence adoption accelerates worldwide, next-generation DDR5 memory is experiencing a critical supply shortage. To avoid delays in AI data center deployments, Meta has partnered with industry leaders like TSMC and Micron to enginee
Related Special Topic Recommendations
Comments (0)
0/500

In the realm of artificial intelligence, open-source models have long served as a vital driver for improving technological accessibility across the globe. Yet recently, the open-source model Rio3.5397B — developed by an IT firm affiliated with Brazil’s Rio de Janeiro municipal government — found itself at the center of a public debate concerning its originality.
The controversy originated from a technical assessment conducted by the Nex-AGI team. Their findings revealed that this model, which had previously been regarded as a dark horse contender, exhibited strong indications of being constructed from existing components. Through thorough mathematical and statistical analysis, the researchers determined that roughly 60% of the model’s core parameters were derived from Nex N2Pro, while the remaining 40% came from
To confirm their hypothesis, the analysis team removed the model’s system prompt and observed surprising results: in subsequent interactions, the model had a 79% chance of identifying itself as “Nex from Nex-AGI” and was able to accurately repeat customized background narratives specific to the Nex product line. Additionally, statistical evaluations showed that the distribution of the model’s network layers followed an exceptionally precise 0.6 to 0.4 ratio, with a statistical deviation of thousands of standard deviations — levels that made coincidence virtually impossible.
In response to this situation, the Nex team reacted with calm rather than anger, noting sarcastically that the other party’s actions actually underscored the superiority and robustness of their own model technology. They also emphasized that while the open-source community promotes sharing and innovation, “reasonable use” must be grounded in fundamental ethical principles such as proper attribution and acknowledgment. Such unauthorized recombination clearly ran counter to the core values of open source.
To date, the involved parties have not issued an official explanation for why Rio3.5 exhibits such precise overlaps in its features. This incident serves as a reminder for the current open-source ecosystem of large AI models: as competition intensifies around performance, ensuring the originality and compliance of these models has become an essential trust challenge for the industry.
Meta Wants Your Next Gadget to Be Muse-Infused
Meta’s Muse, an AI assistant that handles travel bookings, form submissions, and shopping, has gained widespread adoption, yet a new initiative may especially intrigue developers and hardware enthusiasts.On Friday, Meta unveiled Muse Gadgets, an open
Wan-Streamer v0.2 delivers ultra-low latency AI chat
As modern technology advances rapidly, human-AI interaction is becoming more common. Traditional video calls often struggle with lag, delays, and audio-visual desynchronization. Wan-Streamer v0.2, the latest release from Tongyi Lab, overcomes these c
Meta Joins TSMC and Micron to Equip New Servers with Old Memory
As artificial intelligence adoption accelerates worldwide, next-generation DDR5 memory is experiencing a critical supply shortage. To avoid delays in AI data center deployments, Meta has partnered with industry leaders like TSMC and Micron to enginee











