About the tool
Open-source framework for frontier AI evaluations from the UK AI Security Institute.
Inspect is a Python framework for building and running evaluations that measure coding, agentic tasks, reasoning, knowledge, and multimodal understanding. It includes built-in agent implementations, multi-agent primitives, and sandboxed tool execution, and is maintained by the UK AI Security Institute.
- Pricing
- Open source
- Category
- Evals
- Maker
- Not publicly supplied
- Provenance
- Indexed by Attest from public information
✓ Attested
✓ Attested