We’d like to clarify a small documentation inconsistency to avoid any confusion during evaluation. The VibroBraille demo video and initial prototype were built and demonstrated using Gemini 3, which powers the core semantic compression, multimodal understanding, and tactile-optimized content restructuring shown in the demo. During later production-phase testing, we also validated the system with Gemini 1.5 Flash to manage token limits and ensure cost-efficient, stable operation during prolonged use. The GitHub README currently reflects this deployment configuration, and due to an oversight was not updated to explicitly reference Gemini 3 for the demo phase. The system architecture is model-agnostic, and Gemini remains central to VibroBraille’s semantic-to-tactile pipeline in both configurations. We’ve now updated the documentation to clearly reflect this. We appreciate your understanding and thank you for reviewing our project.
Log in or sign up for Devpost to join the conversation.