Published 20 days ago • loading... • Updated 20 days ago
Aureka Releases OpenDDE, an Open-Source Drug Discovery Engine Designed to Accelerate AI-Driven Therapeutic Discovery
The model has 655 million trainable parameters and is released under the Apache-2.0 license for researchers and drug discovery teams.
On Monday, July 6, 2026, AI TechBio company Aureka released Open Drug Discovery Engine , an open-source biomolecular foundation model designed to serve as the structural reasoning core for next-generation drug discovery systems.
Aureka trained the 655 million parameter model over approximately 414,000 GPU-hours, reflecting a broader shift toward addressing biological discovery as an infrastructure problem requiring massive compute and data engineering.
OpenDDE demonstrates strong antibody-antigen co-folding performance, reaching an 81.9% success rate on the 2026ARK-AB benchmark under oracle selection while narrowing the gap with reported IsoDDE-level results.
Releasing training code, inference pipelines, and benchmarks under the Apache-2.0 license, Aureka aims to enable independent validation and global collaboration among researchers, startups, and academic laboratories.
Future development will integrate de novo molecular design and affinity estimation with a high-throughput automated wet-lab platform, creating a closed-loop discovery system for functional antibody development.