LIMI: Less is More for Agency

2025-09-22

AI Summary

Despite strong reasoning and generative capabilities, current AI systems lack agency—the capacity to autonomously execute tasks, manipulate tools, and produce tangible outcomes. Method: Challenging the prevailing “more data yields stronger agents” paradigm, we propose a “less-is-more” approach to cultivating agency: training models on only 78 high-quality, human-annotated demonstrations of autonomous behavior—spanning problem identification, hypothesis formulation, and solution execution—within collaborative software development and scientific research workflows. Contribution/Results: We establish the *Agency Efficiency Principle*, demonstrating that demonstration quality—not dataset scale—is decisive for autonomy. Our method achieves 73.5% on a comprehensive agency benchmark, outperforming state-of-the-art models trained on ~10,000 samples by 53.7% absolute gain, despite using merely 1/128 the data volume. This advances AI from passive “thinking” toward active “acting,” marking a pivotal shift toward agentic intelligence.

Technology Category

• Multiagent Systems: Modeling other Agents
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Agentic search

Abstract

We define Agency as the emergent capacity of AI systems to function as autonomous agents actively discovering problems, formulating hypotheses, and executing solutions through self-directed engagement with environments and tools. This fundamental capability marks the dawn of the Age of AI Agency, driven by a critical industry shift: the urgent need for AI systems that don't just think, but work. While current AI excels at reasoning and generating responses, industries demand autonomous agents that can execute tasks, operate tools, and drive real-world outcomes. As agentic intelligence becomes the defining characteristic separating cognitive systems from productive workers, efficiently cultivating machine autonomy becomes paramount. Current approaches assume that more data yields better agency, following traditional scaling laws from language modeling. We fundamentally challenge this paradigm. LIMI (Less Is More for Intelligent Agency) demonstrates that agency follows radically different development principles. Through strategic focus on collaborative software development and scientific research workflows, we show that sophisticated agentic intelligence can emerge from minimal but strategically curated demonstrations of autonomous behavior. Using only 78 carefully designed training samples, LIMI achieves 73.5% on comprehensive agency benchmarks, dramatically outperforming state-of-the-art models: Kimi-K2-Instruct (24.1%), DeepSeek-V3.1 (11.9%), Qwen3-235B-A22B-Instruct (27.5%), and GLM-4.5 (45.1%). Most strikingly, LIMI demonstrates 53.7% improvement over models trained on 10,000 samples-achieving superior agentic intelligence with 128 times fewer samples. Our findings establish the Agency Efficiency Principle: machine autonomy emerges not from data abundance but from strategic curation of high-quality agentic demonstrations.

Problem

• Challenges the paradigm that more data yields better AI agency
• Demonstrates agency follows different principles than language modeling
• Shows efficient cultivation of machine autonomy requires strategic curation

Innovation

• Achieves agency with minimal curated training samples
• Demonstrates superior performance using only 78 demonstrations
• Establishes agency emerges from quality not quantity