June 2026 – Digital Promise’s K-12 AI Infrastructure Program has launched a Request for Proposals (RFP) for an up to $8 million Gates Foundation-managed grant to build an open-source AI model for K-12 math tutoring in the United States. Known as EDU AI, the initiative was officially released on June 1, 2026, with applications due by July 31, 2026.
The grant will fund one award, with a project period of 30 to 36 months beginning in November 2026. The goal is to create education-specific AI infrastructure that can support one-to-one math tutoring in U.S. classrooms, improving student motivation, engagement, metacognition, and learning outcomes.
The RFP highlights a critical challenge: current AI tutors do not yet behave like effective human tutors. They often provide answers too quickly, miss signs of student motivation, and fail to distinguish between misconceptions and simple arithmetic slips. For math education, these limitations undermine the potential of AI to support productive struggle and deeper learning.
Applicants are expected to deliver open-source outputs such as model weights, training and fine-tuning code, datasets where permissible, evaluation tools, documentation, and reference implementations. All developments must be released under permissive licenses, including Creative Commons Attribution 4.0 for content and Apache 2.0 for software and models.
Eligibility requirements include prior experience with large language models, peer-reviewed publications before May 8, 2026, and a record of contributing digital public goods. Teams must demonstrate meaningful prior deployment or evaluation using real student data and include expertise in AI engineering, K-12 practice, learning science, and EdTech product partnerships. At least one major tutoring EdTech provider must be identified for Phase 3 integration testing.
Safety is a non-negotiable requirement. Proposals must include plans for student data protection, anonymization, compliance with U.S. privacy laws, and bias mitigation strategies. Models will be tested for validity, reliability, fairness, safety, and efficacy, with feedback from teachers and school administrators integrated throughout development.
The project will coordinate with upcoming benchmarks, including the AllenAI and Stanford Scale AI tutoring benchmark, and align with existing initiatives such as the National Tutoring Observatory, the AI Math Tutoring Benchmark and Open Dataset Project, and the Math Misconceptions Data Challenge.
This RFP is part of Digital Promise’s wider $26 million K-12 AI Infrastructure Program, which supports openly shared datasets, models, and benchmarks to address gaps in generative AI for education. By focusing on public goods, the program aims to ensure that AI tools are designed with learning science principles and tailored to the variability of K-12 learners.
The EDU AI initiative represents a major step toward building open-source AI tutoring models that can transform math education, offering scalable, safe, and effective support for students across the United States.
