# Inspect AI

> MIT LLM and agent evaluation framework from the UK AI Security Institute

**Category:** AI Agents  
**Pricing:** open-source  
**URL:** https://inspect.aisi.org.uk  
**GitHub:** https://github.com/UKGovernmentBEIS/inspect_ai  
**Added:** 2026-10-20

---

## The Hook

Inspect AI is the UK AI Security Institute's framework for LLM and agent evaluations, with sandboxed tool use and a log viewer for looking at each run.

## What's Good

MIT (UKGovernmentBEIS/inspect_ai), release 0.3.282 (2026-10-19). ★~3.0k (~19 a week). Useful when you need repeatable evals of models or agents that use tools, and want to inspect each run afterwards.

## The Catch

It is still 0.x and releases near-daily, so pin a version or your eval results can shift under you.

## Verdict

A serious, actively developed eval framework; pin it.