Model Council

Watch the frontier models argue it out

Ask a question and it goes out simultaneously to every frontier flagship model at its highest reasoning setting. Each model then reads the others' answers — anonymised, with no memory of writing its own — and evaluates them. Claude Opus reads every evaluation and delivers a final verdict. Where the council agrees is a strong signal for what AI answer engines will converge on; where it splits is a gap worth writing about. Runs on your own OpenRouter key, in your browser.