Anthropic and Adaptive Bio test five thousand synthetic proteins, experimental verification breaks the isolation of digital models

Anthropic launched a research partnership with Adaptive Bio to organize an open competition in protein engineering and AI-based design, aiming to experimentally validate more than five thousand protein designs inside Adaptive's automated labs without charging participants any fees. This step represents a direct effort to move AI models from closed digital simulation environments to wet-biology testing, where molecules are actually manufactured and their properties examined in a real laboratory setting that determines their functional efficiency.

Anthropic and Adaptive Bio share joint sponsorship of one million dollars for laboratory screening and verification, while Anthropic provides an additional one million dollars in compute credits for its Cloud model to the selected teams and researchers.The system is completed by a contribution from Modal, which provides additional cloud compute capacity to support operations, together with Twist Bio Science supplying the DNA required for laboratory testing, giving contestants an integrated supply chain that starts with code, passes through DNA, and reaches automated synthesis.

The competition opens for Cloud support applications between 16 and 24 September 2026, with accepted researchers notified before the challenges officially launch on 28 September. The contest runs until 31 October 2026, introducing a new challenge each week, and includes an open track that allows contestants to use any design method they choose without requiring Cloud or providing support credits, while the organizers acknowledged the absence of direct cash prizes, allocating non-monetary in-kind awards to be announced later.

Laboratory screening and verification stages for all selected designs will be completed by 30 November 2026, before the full results and experimental data are publicly released on the ProteinBase platform on 15 December 2026.This open publication provides an experimental reference that enables measurement of computational prediction algorithm accuracy against actual biological behavior inside the automated lab.

For researchers and biotech teams in the Gulf, Egypt, and the Levant, this development resolves a central dilemma that hinders regional drug discovery and enzyme design: the chronic gap between possessing computational capabilities to design molecules and lacking automated wet labs that can synthesize and test rapidly without massive capital costs. Opening a free laboratory testing window for five thousand designs and releasing the full data in December gives Arab researchers the opportunity to test their hypotheses and compete with global teams without needing a complex laboratory budget. If you lead a research team or build a biotech startup, your practical step is to take advantage of the rapid registration cycles to test your designs in the lab, or to prepare to engineer your models on the open database as soon as it becomes available at year-end.