AI Capabilities
Comprehensive tracking of AI progress across key capability dimensions that contribute to artificial general intelligence
Capability Breakdown
Language
Weight: 10%Natural language understanding and generation across multiple languages and styles. Includes nuanced communication and contextual awareness.
Reasoning
Weight: 25%Logical thinking, problem-solving, and abstract reasoning capabilities. Includes mathematical reasoning, logical deduction, and complex multi-step problem solving.
Coding
Weight: 15%Software development capabilities including code generation, debugging, and understanding complex codebases. Encompasses multiple programming languages and paradigms.
Multimodal
Weight: 10%Processing and generating content across multiple modalities including text, images, audio, and video. Integration of different input types for comprehensive understanding.
Science
Weight: 10%Scientific reasoning, hypothesis generation, and research assistance. Encompasses understanding of scientific literature and experimental design.
Agency
Weight: 20%Autonomous action-taking, tool use, and multi-step task completion. The ability to plan, execute, and adapt strategies in complex environments.
Robotics
Weight: 10%Physical world interaction through robotic systems. Includes manipulation, navigation, and embodied reasoning in real environments.
How We Measure Capabilities
Our capability scores are derived from benchmark results, expert assessments, and real-world performance metrics across standardized tasks.
Read our methodology