The Eureka Machine is a system that automates the scientific method, developed by Recursive Superintelligence. It automates the full research loop: proposing an idea, implementing it, running an experiment and validating the result, then using what it learns to choose the next experiment.
It runs many research threads in parallel over long horizons, keeps useful context from prior experiments, combines promising branches, and puts results through validation for reward hacks and variance before treating improved performance as real progress. It builds on principles of open-ended algorithms and recursively self-improving AI.
Recursive applies the Eureka Machine first to the full frontier AI R&D pipeline—from data and training recipes to infrastructure and performance optimization. Early results (June 2026) reached state-of-the-art on the NanoChat Autoresearch, NanoGPT Speedrun and SOL-ExecBench (GPU kernel optimization) benchmarks. The system was also used to build the Flint coding model (MoE, 40B parameters, ~2B active).
