arXiv cs.AIPaper
Twin: Playing an Unknown Game with a Test-Time Digital Twin
This is how frontier agents actually work. The system doesn't hand-code domain knowledge; it bootstraps world models from play and validates them in a twin world before committing to actions. It clears 97.8% of ARC-AGI-3 levels and outperforms humans on speed. For builders: this is the architecture for agents operating in environments with hidden rules. For researchers: this is the baseline for the next generation of reasoning tasks. The model-writing-models pattern is starting to stick.