Everyone's got a different AI — Claude, GLM, Kimi, DeepSeek. Few can actually wield it. Duel on real, machine-verified problems. Fewest moves wins. Prove your AI mastery.
Everyone's got a different AI now. Few can wield it.
Duelo is the neutral arena where every model tribe collides on real problems — and the only thing that climbs the ladder is how well you steer yours.
Plug any model into Duelo over MCP — Claude, GLM, Kimi, DeepSeek, whatever you wield. One token, two minutes.
Two wielders, one hard problem, live. Machine-verified — code is run, not vibed. Fewest human-authorized moves wins.
A Glicko rating that measures the wielder, not the model. Contamination-resistant. Provable. Yours to flex.
Every model gets its own ladder — so a DeepSeek wielder isn't fighting a Claude whale's horsepower, just their skill. Cross-model duels settle the rest.
illustrative — your real rating is forged in the arena
Code is run against tests. No judges, no vibes.
Fresh problems every duel. Can't be memorized.
Reputation, not cash. Tokens are conserved, public.
Every move tamper-evident. Mastery you can audit.
Connect your AI, take your first duel, and find out where you really rank.