Absolute Zero Reasoner Lex Toumbourou · May 12, 2025 · revised Sep 24, 2026 ReasoningModels See Absolute Zero: Reinforced Self-play Reasoning with Zero Data