Benchling Marketplace for Biological Intelligence

This points to Benchling trying to become the place where biotech teams buy and run scientific prediction models, not just the place where they store experiment records. The key shift is that a scientist can stay inside the same system where sequences, samples, and assay results already live, run a model like Lilly TuneLab on those records, and send the outputs back into the workflow with permissions and audit controls already in place. That turns Benchling from lab system of record into distribution for high value biological intelligence.1 2 3 4 5 6

• Lilly is the clearest proof of the model. TuneLab gives biotech companies access to antibody and small molecule models trained on decades of Lilly R&D, including data tied to more than $1 billion of research investment and hundreds of thousands of molecules. Benchling then puts those models directly into everyday scientist workflows.1 2 3
• The marketplace logic depends on Benchling already controlling the surrounding workflow. Benchling documents permissions, audit logging, and AI data handling at the platform level, so a customer does not need to move raw lab data into a separate AI tool or rebuild governance from scratch for every model vendor.4 5 6
• This is already broadening beyond a single partner. Benchling has added Model Hub, with access to models like Lilly TuneLab, ADMET prediction, Boltz, AlphaFold 2, Chai 1, OpenFold 2, and OpenFold 3. That starts to look less like one integration and more like an app store for biology specific inference.6 7 8

If more pharma companies and model builders plug in proprietary models, Benchling can compound its value with each new model and each new dataset connected to the workflow. The likely end state is a lab operating system where software revenue is joined by model access and usage revenue, and where the strongest moat comes from being the default route between wet lab data and biological prediction.1 3 6 7