Search epoch.ai
Data on AI
Epoch AI collects key data on machine learning models from 1950 to the present to analyze historical and contemporary progress in AI. Our database is a valuable resource for policymakers, researchers, and stakeholders to foster responsible AI development and deployment.
Explore our data
Notable AI Models
Our flagship dataset, tracking AI progress from 1950 to today, with 800+ notable models and 400+ training compute estimates.
Updated January 16, 2025
Large-Scale AI Models
Data on 200+ AI models trained with more than 1023 floating point operations, at the leading edge of scale and capabilities.
Updated January 16, 2025
Machine Learning Hardware
Key data on 130+ AI accelerators, such as GPUs and TPUs, used for developing and deploying ML models in the deep learning era.
Updated January 16, 2025
AI Benchmarking Hub
Track the performance of leading AI models on challenging benchmarks, with insights into compute and accessibility.
Updated November 27, 2024
Use our work
Epoch AI’s data is free to use, distribute, and reproduce provided the source and authors are credited under the Creative Commons Attribution license. Citations can be found on the respective pages for each dataset.
Download our data
Notable AI Models
CSV, Updated January 16, 2025
Large-Scale AI Models
CSV, Updated January 16, 2025
Machine Learning Hardware
CSV, Updated January 16, 2025
AI Benchmarking Hub
CSV, Updated November 27, 2024
Our methodology
We identify and track contemporary and historic advances in AI, collating key details across several areas. This research includes who developed models, when, and for what tasks, how much compute was used for training, how many parameters models have, how much data was used for training, what hardware was used for training, and more.
Our research
Epoch AI is a multidisciplinary research institute investigating the trajectory and impact of artificial intelligence.
We publish datasets, data visualizations, research reports, and predictive models to analyze the forces shaping AI development. Our goal is to foster scientific dialogue and bring empirical rigor to predictions about the future of AI.
