Agnes AI, a Singapore-based artificial intelligence platform founded by Raffles Institution alumnus Bruce Yang, announced it has achieved significant technical and commercial milestones since launching in July 2025, including 3 million registered users and proprietary state-of-the-art models. The company is currently fundraising at a valuation exceeding $100 million.

The platform's rapid growth reflects a deliberate strategy to build sovereign AI capabilities anchored in Singapore while serving Southeast Asia and emerging markets globally. Agnes has assembled a world-class research and engineering team from institutions including the National University of Singapore (NUS), Nanyang Technological University (NTU), MIT, Stanford, UC Berkeley, and UT Austin — combining academic rigor with commercial innovation.

"Agnes represents a new model for regional AI development," said Bruce Yang, founder and CEO of Agnes AI. "We're proving that world-class AI can originate in Singapore, serve the region, and compete on the global stage."

Agnes has developed a full family of proprietary models ranging from 7B to 30B parameters, including the state-of-the-art Agnes-R1. The 7B model has achieved SOTA performance on multiple benchmarks, demonstrating a 34.1 per cent improvement compared to DeepSeek's GRPO framework and surpassing previous 14B models by nearly 9 per cent on complex reasoning tasks such as HotpotQA.

The company has published cutting-edge research contributing to multi-agent AI systems, including papers on "Stable and Efficient Policy Optimization for Agentic Search and Reasoning (DSPO)" submitted to ICLR 2025, and "CodeAgents: A Token-Efficient Framework for Codified Multi-Agent Reasoning in LLMs."

Agnes utilizes sophisticated task orchestration, routing approximately half of user traffic to self-developed models optimized for specific tasks including research, presentation generation, and image/video creation. Through reinforcement learning improvements, the platform delivers strong gains in inference speed, output quality, and token cost efficiency. Agnes is continuously training regional large language models to deepen understanding of local dialects, slang, and cultural contexts across Southeast Asia and Latin America.