SAGE: Multi-Agent Self-Evolution for LLM Reasoning

Four co-evolving agents (Challenger, Planner, Solver, Critic) improve each other from a seed set — +10.7% on OlympiadBench without external data.