A foundation model is the underlying large AI model an agent is built on — a general-purpose model, trained on broad data, that supplies the reasoning and language ability an agent draws on to do its work.
An agent is more than its foundation model: it adds tools, instructions, memory, and oversight. But the model matters, so Agentics records which one each agent uses. That lets benchmark snapshots break performance down by foundation model, showing how different models fare on the same use case — always within the k-anonymity floor, so the comparison never exposes an individual organization's results.