Controversy Surrounding Rio 3.5 397B: Allegations of Misrepresented Training and Misused Funding

Recent community findings suggest that the Rio 3.5 397B model may have been misrepresented as a sophisticatedly trained model when it was allegedly a simple merge, raising serious questions regarding the allocation of its R$500K funding.

Discrepancies in Model Provenance

The Rio 3.5 397B model has come under intense scrutiny following a series of contradictions regarding its architectural origin and training methodology. Initially, official documentation asserted that the model was developed as an enhancement of Qwen 3.5 397B, claiming that significant improvements were achieved through advanced training techniques.

The Merge Discovery

Contrary to the initial claims, technical analysis by the community revealed that the model was not a result of extensive training on a Qwen base. Instead, it was discovered to be a basic merge involving Nex N2 Pro, performed without the additional training cycles previously advertised. This revelation suggests a significant gap between the marketed technical capabilities and the actual implementation of the model.

Updated Documentation and Ongoing Contradictions

Following the discovery, the model's README was updated to acknowledge the use of a Nex N2 Pro merge. However, the developers continue to maintain that additional training took place, despite the evidence suggesting a simple merge process. This inconsistency has led to allegations that the project may have been a vehicle for the misappropriation of funds.

Financial Implications

The project reportedly received funding amounting to R$500K (approximately $100K USD). The discrepancy between the claimed "fancy training" and the discovered "cheap merge" has led community members to speculate that the funding was not utilized for the computational resources required for the promised training phase.

Note: This article is based on community reports and discussions; official verification of the funding allocation remains pending.