AI and models · Independent · 21 Oct 2026

UKGovernmentBEIS/inspect_ai

Inspect: A framework for large language model evaluations

← AI and models
Inspect: LLM evaluation framework. Tests large language models with automated tasks and scores.

In simple English

Language models can say many things, but knowing if they are safe or correct is hard. This tool gives you a way to test them with hundreds of ready-made challenges. You write tasks, send questions to the model, and grade the answers. It comes from the United Kingdom Artificial Intelligence Safety Institute. You need some Python skill to run it and time to set up your test sets.

What it is

Inspect is a Python framework for large language model (a computer program trained on many words to generate text) evaluations, created by the UK AI Security Institute. It includes over two hundred pre-built tests, a command-line interface (a text-based way to run programs), and a web viewer to check results.

What it is good for

Solo builders who want to test if a new language model prompt, fine-tuned model, or agent can solve specific coding, math, or reasoning problems before putting it in front of users.

Stars3.1k
LicenceMIT
Latest0.3.282 (2026-10-19)checked 21 Oct 2026
Last code change2026-10-21
Written inPython
Setup effort
3 / 5
Payoff
4 / 5
Good
  • Includes over two hundred pre-built evaluations you can run right away.
  • Supports multi-turn conversations and model-graded scoring.
  • Comes with an MIT licence that allows free commercial use.
Watch for
  • Requires writing Python code to build custom evaluations.
  • Needs access to language model API keys which can cost money per test run.
  • Has many open issues that may show rough edges in niche features.
Replaces

manual prompt testing in chat windows

Works well with

uv

Similar repositories
  • vibrantlabsai/ragas16k stars

    Evaluates retrieval-augmented generation pipelines, whereas Inspect tests general language model tasks and safety.

  • EleutherAI/lm-evaluation-harness14k stars

    Runs academic benchmarks on language models, whereas Inspect focuses on programmable task evaluations and tool use.

Also reviewed on Repo Radar
  • ggml-org/llama.cpp132k stars

    A C and C++ library and command-line tool for running language models locally.

  • EleutherAI/lm-evaluation-harness14k stars

    Runs standard question sets against a language model and reports the scores.

  • mudler/LocalAI50k stars

    Serves text, vision, speech, image and video models from one self-hosted service that imitates the paid interfaces.

Install
git clone https://github.com/UKGovernmentBEIS/inspect_ai.git
cd inspect_ai
pip install -e ".[dev]"

Independent: reviewed for Repo Radar outside the daily editions. Every independent repository

On Stack Radar: AI & Automation › AI Development & Evaluation › Model Evaluation.

Get the next edition in your inbox

A dozen repositories, opened and checked. The licence read, the last release dated, and the ones that did not make it named with the reason. It is the half most lists leave out.

No tracking pixels. One click to leave. The archive stays free either way.

We use your address to send the edition and nothing else. Confirm by email, leave in one click. How we handle it.