Deadline Date: May 27, 2026
Innovate UK has launched a funding competition to support the development of high-quality benchmarking datasets that enable robust evaluation and advancement of frontier AI systems.
The aim of this competition is to support projects that create high-quality benchmarks based on representative dataset slices to evaluate new AI models, while also developing larger, AI-ready annotated and curated datasets, with solutions aligned to AI-Enabled Health and Life Sciences and Advanced Materials with AI, focusing on applications such as medicines discovery, predictive healthcare, clinical trials, aerospace, net zero technologies, defence materials, and semiconductors.
This funding opportunity, backed by a minimum investment of £4.5 million, seeks to strengthen the UK’s position in cutting-edge AI development by enabling the creation of reliable and comparable benchmarking frameworks. Projects are expected to go beyond existing datasets by demonstrating clear added value in improving the validation and performance assessment of AI and machine learning models.
Applicants are encouraged to form consortia that combine data-owning organisations with partners skilled in data engineering, annotation, and benchmarking. The collaboration should ensure that datasets are not only technically robust but also practically useful for industry adoption and future development.
Each proposal must deliver an open benchmark package that includes a clearly defined task, evaluation protocol, accessible dataset, and a complete evaluation harness usable by third parties, along with proper documentation and metadata. In addition, projects must produce a fully curated and annotated dataset and clearly outline intellectual property licensing and access routes for dataset utilisation.
For projects involving health and life sciences data, strict data governance and privacy protections are mandatory, ensuring that any released data is anonymised or de-identified. This ensures compliance while still enabling meaningful AI development.
Eligible projects must have total costs between £500,000 and £750,000, run for a duration of 6 to 12 months, and begin by 1 September 2026. All funded activities must be conducted within the UK, and applicants must demonstrate intent to exploit project outcomes within the country.
To lead a project, the applicant must be a UK-registered business or research and technology organisation, and the consortium must include at least one SME. Academic institutions may participate as collaborators but are not permitted to lead applications.
This initiative represents a targeted effort to build high-quality AI infrastructure, supporting both innovation and commercialisation while ensuring that datasets and benchmarks are accessible, reliable, and aligned with national priorities.
For more information, visit GOV.UK.
