The decision-making powerhouse, Decitron, launched by Wenge AI, has garnered acclaim as the world's inaugural general-purpose decision-making large model. This innovative model seamlessly amalgamates functionalities such as world modeling, multi-agent systems, and game-theoretic problem-solving into a cohesive decision-making framework tailored for the open world, thereby establishing a fully integrated closed loop. Its foundational technologies encompass the explicit world model MetaWorld, SAO formal representation, and the hybrid reasoning architecture AutoABM, which collectively facilitate end-to-end decision-making inference, spanning from comprehending the world to appraising future scenarios.

Take trade conflict simulation as an illustration. Initially, the model meticulously organizes and filters information, converting raw facts into perpetually updated world states. Subsequently, it autonomously constructs a multi-agent simulation environment utilizing AutoABM for path simulation, integrates formal problem-solving techniques to ascertain strategy viability, and culminates in furnishing probabilistic future trajectories through predictive analysis and calibration.

Experimental evidence underscores that this decision-making framework surpasses several leading large models in short-term event forecasting and open prediction market benchmark assessments. Decision-making AI, which centers its computational efforts on alterations in world states, propels the competitive landscape in AI from individual models to comprehensive systems, thereby delineating a technological roadmap for general-purpose decision-making large models.