IplanRIO, an IT company under the Rio de Janeiro city government, apologized on HuggingFace after its open-source model Rio-3.5-Open-397B was accused of being a mixed-weights version of other models rather than an independently trained system.
According to Odaily, the model gained rapid attention globally before the narrative shifted within 24 hours, with allegations that it was a repackaged combination of Alibaba’s Qwen3.5 and NexN2Pro weights.
IplanRIO said it mistakenly uploaded a merged baseline version intended for comparison.
The open-source model team Nex-AGI publicly alleged that the Rio model showed no signs of being trained by Rio and claimed it was a weights mixture of NexN2Pro and Qwen3.5-397B-A17B.
AI TRENDS | IplanRIO Apologizes After Rio-3.5-Open-397B Faces Claims of Mixed Weights
2026-06-15 10:04:02
Disclaimer:
1. The information provided does not constitute investment advice. Investors should make independent decisions and bear all risks themselves.
2. The copyright of this content belongs to the original author. The views expressed herein are solely those of the author and do not represent the stance or position of this website.
Previous article：
Birkenstock Plans First Bond Sale in Over Five Years
