Allen Institute for AI
Allen Institute for AI (Ai2) is a non-profit AI research institute that develops foundational AI research and innovation to deliver real-world impact through large-scale open models, data, robotics, conservation, and scientific discovery.
Founding Story
The Allen Institute for AI was founded in 2014 by Microsoft co-founder and philanthropist Paul Allen, who appointed Oren Etzioni in September 2013 to direct research prior to the formal launch. It was established as one of three Allen Institutes (including Brain Science and Cell Science) with a mission to conduct high-impact AI research and engineering for the common good. The institute was created to contribute to humanity through breakthrough AI research focused on truly open AI development.
Discussions
No discussions yet
Be the first to start a discussion about Allen Institute for AI
Leadership
Founders
Paul Allen
Co-founder of Microsoft. Philanthropist who created multiple Allen Institutes including Brain Science and Cell Science. Passed away in 2018.
Oren Etzioni
Founding CEO (2013-2022). PhD in computer science, former professor at the University of Washington. Founded or co-founded several companies including Farecast, Decide, MetaCrawler, and Netbot. Appointed in September 2013 to direct research prior to the formal launch.
Executive Team
Ali Farhadi
Chief Executive Officer
Former Chief of Machine Learning at Apple. Previously Associate Professor at University of Washington. Co-founder of XNOR.ai which was acquired by Apple. Named to AI Power List in 2026 for open-source AI focus. Appointed CEO in June 2023.
Noah Smith
Senior Director of Natural Language Processing Research; OMAI Project Leader
Computer science professor at the University of Washington. Leading the NSF-funded Open Multimodal AI Infrastructure to Accelerate Science (OMAI) project.
Business Model
Revenue Model
Non-profit research institute primarily funded by grants and philanthropic donations. Founded with endowment from Paul Allen's estate. Generates no direct revenue from products as all AI models, tools, and datasets are provided free and open-source. The institute also operates AI2 Incubator which has spun out as an independent organization making venture investments in AI startups.
Pricing Tiers
All AI models (OLMo, Molmo, Tulu), datasets (Dolma), tools (AllenNLP, AI2-THOR, Asta), and products (Semantic Scholar, EarthRanger, Skylight) are provided completely free and open-source to researchers, developers, and users worldwide.
Target Markets
- Academic researchers and scientists
- AI/ML researchers and developers
- Educational institutions and universities
- Wildlife conservation organizations
- Environmental protection agencies
- Ocean conservation groups
- Scientific research and literature discovery
- Open-source AI model development and research
- Language model training and experimentation
- Robotics and embodied AI development
- Climate science and environmental monitoring
- Wildlife conservation and anti-poaching
- Global research community
- Wildlife conservation organizations using EarthRanger
- Under-resourced countries using Skylight for illegal fishing detection
- Climate science researchers
History & Milestones
Selected by NSF and NVIDIA to lead $152M Open Multimodal AI Infrastructure to Accelerate Science (OMAI) project; Opened new headquarters in Seattle's first major mass-timber office building
Release of OLMo 3 model family and launch of OlmoEarth Platform
Launch of Molmo 2 for open video understanding and announcement of OLMo 3.1
Open-sourcing of OLMo 1B and 7B variants
Release of OLMo 2 (7B and 13B) and Tulu 3
5 AI Tools by Allen Institute for AI
olmOCR
11dolmOCR is an open-source toolkit by AI2 for converting PDFs and document images into clean, structured plain text using vision-language models.
AutoDiscovery uses Bayesian surprise to autonomously explore datasets and uncover surprising, assumption-challenging insights hidden in your data.
ASTA
2moAI-powered tool for synthesizing and analyzing scientific literature to accelerate research discovery.
SciArena
4moOpen evaluation platform from the Allen Institute for AI where researchers compare and rank foundation models on scientific literature tasks using head-to-head, literature-grounded responses.
OLMo
9moFully open language model family with complete access to training data, code, and weights
