Foundation models for
computer-use agents
We’re OpenAGI Foundation. An open research lab formed by world-class researchers and engineers from MIT, UIUC and CMU. We're building agentic foundation models by redefining pre-training, mid-training, post-training, and large-scale reinforcement learning - to push computers into their Fourth Age: a computer that thinks and acts for you.
We release our work as open source and grow an ecosystem developers can use and extend.
Foundation models for
computer-use agents
We’re OpenAGI Foundation. An open research lab formed by world-class researchers and engineers from MIT, UIUC and CMU. We're building agentic foundation models by redefining pre-training, mid-training, post-training, and large-scale reinforcement learning - to push computers into their Fourth Age: a computer that thinks and acts for you.
We release our work as open source and grow an ecosystem developers can use and extend.
OUR MISSION
What we're building
Agentic Foundation Models
Agentic Foundation Models
The existing agent models inherit the training paradigm that was discovered 3-7 years ago, which was primarily for chat and reasoning (math and coding) scenarios. In the agent era, we need to rethink each stage of such an old paradigm. Agentic foundation models, purposely built for universally-capable agents, redefine the approach of pre-training, mid-training, post-training and large-scale reinforcement learning. The first gigantic use case of agentic foundation models is to push computers into their Fourth Age.
The existing agent models inherit the training paradigm that was discovered 3-7 years ago, which was primarily for chat and reasoning (math and coding) scenarios. In the agent era, we need to rethink each stage of such an old paradigm. Agentic foundation models, purposely built for universally-capable agents, redefine the approach of pre-training, mid-training, post-training and large-scale reinforcement learning. The first gigantic use case of agentic foundation models is to push computers into their Fourth Age.
The Fourth Age of Computers
The Fourth Age of Computers
Computers have always moved in ages.
The First Age: Mainframes. Machines owned us. Only the elite could touch them. Computing was centralized, rigid, inaccessible.
The Second Age: PCs. We shared control. A keyboard, a mouse, a screen — our bridge to the machine. Powerful, but still on its terms.
The Third Age: Smartphones. Computers shrank into our pockets. Always with us, always demanding us. A constant companion, but also a constant distraction.
The Fourth Age: OpenAGI. Now we arrive at the threshold of something new: A computer that thinks and acts for you. The Fourth Age is not a tool. Not a device. Not an app. It is a new definition of computing itself.
Computers have always moved in ages.
The First Age: Mainframes. Machines owned us. Only the elite could touch them. Computing was centralized, rigid, inaccessible.
The Second Age: PCs. We shared control. A keyboard, a mouse, a screen — our bridge to the machine. Powerful, but still on its terms.
The Third Age: Smartphones. Computers shrank into our pockets. Always with us, always demanding us. A constant companion, but also a constant distraction.
The Fourth Age: OpenAGI. Now we arrive at the threshold of something new: A computer that thinks and acts for you. The Fourth Age is not a tool. Not a device. Not an app. It is a new definition of computing itself.
An Open-Source Ecosystem
An Open-Source Ecosystem
We believe every major technical advancement in human history should be open-sourced. By fostering a collaborative environment, we can accelerate innovation and ensure that the benefits of technology are shared by all. We want to democratize and bring super intelligence to your own hands, so that it shines thousand times more brightly.
We believe every major technical advancement in human history should be open-sourced. By fostering a collaborative environment, we can accelerate innovation and ensure that the benefits of technology are shared by all. We want to democratize and bring super intelligence to your own hands, so that it shines thousand times more brightly.
WE'RE HIRING
Join our lab
Join our world-class team to design, train, and evaluate agentic foundation models, grow an open-source ecosystem, and build products that people love.
Join our world-class team to design, train, and evaluate agentic foundation models, grow an open-source ecosystem, and build products that people love.
Open Positions (5)
Open Source Community & Developer Relations Lead
Open Source Community & Developer Relations Lead
AI Agent Backend / Research Engineer
AI Agent Backend / Research Engineer
Research Scientist (LLMs, Agents)
Research Scientist (LLMs, Agents)
Research Engineers (ML/AI)
Research Engineers (ML/AI)
Operation / Marketing / Digital Content Specialist
Operation / Marketing / Digital Content Specialist
GET INVOLVED
Join our community
As part of our selective early community, you’ll influence the SDK, share feedback loops, and help define computer use in the AI era.
As part of our selective early community, you’ll influence the SDK, share feedback loops, and help define computer use in the AI era.
2025 OpenAGI Research Foundation
© All Rights Reserved