Rio-3.5-Open-397B ≈ 0.6 x Nex-N2_pro + 0.4 x Qwen #4
Summary
This article discusses the Rio-3.5-Open-397B model, claimed to be a 0.6 Nex-N2_pro and 0.4 Qwen blend. It argues the model is not independently trained and provides evidence from prompts and weight tensors across 60 layers. The post raises provenance and attribution concerns for open-source AI models.