Skip to main content

OpenAGI

Category: AI Agents

OpenAGI develops Lux, an advanced AI agent that uses computer screenshots to perform operations in desktop applications autonomously OpenAGI was founded in 2024. The company is led by Zengyi Qin. Based in San Francisco, California, United States. Team size: 11-50. Total funding raised: Unverified (company emerged from stealth in late 2024). Latest round: Seed stage (exact details undisclosed). Key investors include N/A.

Founded
2024
Headquarters
San Francisco, California, United States
Team size
11-50
Total funding
Unverified (company emerged from stealth in late 2024)

Value proposition

Lux claims superior performance over OpenAI and Anthropic models with 83.6% success rate on Online-Mind2Web benchmark at one-tenth the cost

Products and solutions

Lux AI Agent, Agentic Active Pre-training Technology, Computer-Use Foundation Models

Unique value

First foundation model built specifically for computer use, trained with Agentic Active Pre-training using real-world computer interaction data

Target customer

Enterprises and developers requiring autonomous AI agents for complex computer-mediated tasks

Industries served

Technology, Enterprise Software, AI Research, Automation

Technology advantage

Agentic Active Pre-training methodology that learns by acting in digital environments rather than passive text ingestion, achieving 10x cost efficiency

How they differentiate

Claims 83.6% success rate on Online-Mind2Web benchmark, outperforming competitors while costing 10x less and executing actions in 1 second

Main competitors

OpenAI Operator, Anthropic Claude, Google Gemini CUA

Key partnerships

MIT, CMU, UIUC research collaborations

Notable customers

N/A

Major milestones

December 2024: Emerged from stealth mode, December 2024: Released Lux AI agent, December 2024: Claimed performance superiority over OpenAI and Anthropic

Growth metrics

Claims 83.6% success rate on Online-Mind2Web benchmark, 10x cost efficiency vs competitors

Market positioning

Early-stage disruptor in computer-use AI agents with performance-centric positioning

Geographic focus

Primarily US-based with global deployment potential

Patents and IP

No publicly listed patents as of December 2025

About Zengyi Qin

MIT PhD candidate in Computer Science, previously Co-Founder and AI Research Lead at MyShell, expertise in multimodal AI and computer vision

Official website: