This initiative funds projects developing frontier AI models that achieve state-of-the-art performance with multimodal and agentic capabilities such as tool use and autonomous problem-solving. It supports research to reduce computational costs through compression, distillation, and innovative architectures, while ensuring AI safety by design. Projects must address energy efficiency, computational constraints, and safety challenges including misalignment and adversarial robustness, contributing to strengthened European AI capabilities.
Eligible applicants are legal entities established in EU Member States, Iceland, Norway, and specified associated countries including Canada, Israel, Republic of Korea, New Zealand, Switzerland, and the UK. Participation excludes entities controlled by non-eligible countries, especially China, unles
