Real-time streamingWatch every AI write its answer live via WebSocket. No waiting — the debate unfolds as it happens.
Semantic scoringA live agreement score computed via vector embeddings shows how aligned the models are — updated after every message.
Upload documentsAttach PDFs, DOCX, spreadsheets or images. Claude Vision processes images; text is extracted and sent to all models.
Voice inputDictate your question via Deepgram's nova-2 model. Your words stream into the input in real time.
7 debate modesAlethe, Friendly, Business, Wellness, Socratic, Debate, Custom. Each mode changes how every model responds to your question.
Export & shareExport any debate as PDF. Publish a debate publicly and share the link — anyone can read or fork it.