Self-Consistency Engine
A multi-model AI reasoning platform that compares responses from multiple LLMs and uses an AI judge to select the most accurate answer.
React.js
TypeScript
Vite
Tailwind CSS
Node.js
Express.js
Google Gemini API
OpenAI API
Features
- Submit a single prompt to multiple AI models simultaneously
- Compare responses from Gemini 2.5 Flash Lite, Gemma 4 31B, and OpenAI o3-mini
- Use an AI judge model to evaluate and select the most accurate response
- View model responses progressively as they are completed
- Visualize the flow from individual model responses to the final judgement
- Clean and responsive interface built for focused AI experimentation
- Dark mode support with smooth theme transitions
- Backend health-check endpoint for service monitoring
- Validated API requests with clear error handling