General Reasoning - Solve Complex Problems with AI

General Reasoning is an AI research company training models that work reliably over long horizons in non-stationary environments. Their focus spans reinforcement learning, multi-agent systems, and environment design to solve complex, real-world problems. The team, formerly at Meta, develops models that can coordinate and communicate effectively on civilization-scale challenges, moving beyond procedurally excellent but unreliable current LLMs.

Training AI models for long-term planning and adaptation

Building multi-agent systems that coordinate effectively

Developing AI for real-world non-stationary environments

Benchmarking model reliability over extended time horizons

Research on civilization-scale AI problems