UK AI Security Institute

license

MIT(OSI)

source

public(full framework, 200+ evals)

core features withheld

no

The whole framework is released under the MIT license and public, with no proprietary core.

• https://api.github.com/repos/UKGovernmentBEIS/inspect_ai/license recorded 2026-08-13

  GitHub's license endpoint reports spdx_id MIT for UKGovernmentBEIS/inspect_ai and returns the MIT text as the LICENSE body.

• https://github.com/UKGovernmentBEIS/inspect_ai recorded 2026-06-04

  MIT license; framework for LLM evals with 200+ pre-built evals; UK AISI maintainer; 2.2k stars, 5,886 commits

• https://inspect.aisi.org.uk/ recorded 2026-09-29

  Docs home says Inspect is built by the "UK AI Security Institute" and "Meridian Labs" and ships "A collection of over 200 pre-built evaluations ready to run on any model".

• https://pypi.org/pypi/inspect-ai/json recorded 2026-09-29

  The MIT-licensed package description says "Inspect provides many built-in components, including facilities for prompt engineering, tool usage, multi-turn dialog, and model graded evaluations" and that the evals collection is "ready to run on any model". Searched the whole JSON body for pricing, enterprise, license key, subscription and paid: zero matches.

• https://raw.githubusercontent.com/UKGovernmentBEIS/inspect_ai/main/README.md recorded 2026-08-13

  README is a build-from-source guide - clone the repo, `pip install -e ".[dev]"`, build the React web UI from the checked-in submodule, render the docs locally. It names UK AISI as the creator and points at 'over 200 pre-built evaluations ready to run on any model'. There is no pricing, no hosted tier, no enterprise edition and no license key anywhere in it.

Inspect is UK AISI's open-sourced evaluation platform, reported to be used by other governments and some of the leading AI developers. PyPI downloads are high but inflated by continuous-integration and dependency installs, so lab adoption carries more weight, and actual users remain a small population of eval engineers rather than a mass-market base.

• https://inspect.aisi.org.uk/ recorded 2026-06-04

  framework for frontier AI evaluations by UK AISI + Meridian Labs; adopted by major labs

• https://pepy.tech/projects/inspect-ai recorded 2026-06-04

  87.89M total / 9,856,617 last-30-day PyPI downloads (corroboration)

• https://pypistats.org/api/packages/inspect-ai/recent recorded 2026-10-08

  last_month = 2,896,739 downloads of the inspect-ai package (last_week 994,166, last_day 155,525)

• https://www.aisi.gov.uk/blog/our-first-year recorded 2026-10-08

  'we open sourced our evaluations platform, Inspect ... now it is used by other governments and some of the leading AI developers'. The post reports evaluations on 16 models in AISI's first year, and carries no 'thousands of evaluations' figure.

Inspect runs evaluations through one interface over lab APIs, cloud APIs and local servers, and its models page also lists DeepSeek, Moonshot, Perplexity, SageMaker, Fireworks, SambaNova, SGLang and TransformerLens. First-class support for agent and tool-use evaluation and more than 200 built-in evaluations put it among the leaders of the category.

• https://github.com/UKGovernmentBEIS/inspect_ai recorded 2026-06-04

  200+ pre-built evals; tool use, multi-turn, model-graded eval components

• https://inspect.aisi.org.uk/ recorded 2026-09-29

  Docs home says "Agents" combine "planning, memory, and tool use for longer-horizon tasks" and lists "over 20 model providers".

• https://inspect.aisi.org.uk/agents.html recorded 2026-09-29

  Agents page says "Inspect supports a variety of approaches to agent evaluations, including" the built-in "ReAct Agent" and the "Deep Agent" for long-horizon tasks.

• https://inspect.aisi.org.uk/models.html recorded 2026-09-29

  Models page tabulates providers by class, with cloud APIs "AWS Bedrock", "AWS SageMaker" and "Azure AI", hosted open models and local backends "vLLM", "SGLang", "TransformerLens", and says "If the provider you are using is not listed above, you may still be able to use it if" it is added as an extension.

• https://inspect.aisi.org.uk/models.html recorded 2026-09-29

  Models page groups backends into Lab APIs ("OpenAI", "Anthropic", "Google", "Grok", "DeepSeek", "Moonshot AI", "Perplexity"), Open (Hosted) ("Groq", "Together AI", "Fireworks AI", "Cloudflare", "SambaNova") and local ones; it says "If the provider you are using is not listed above, you may still be able to use it if" an extension supplies it.

• https://pypi.org/pypi/inspect-ai/json recorded 2026-09-29

  The package description says Inspect also includes "a collection of over 200 pre-built evaluations ready to run on any model".

• https://raw.githubusercontent.com/UKGovernmentBEIS/inspect_ai/main/README.md recorded 2026-08-13

  README: built-in components for prompt engineering, tool usage, multi-turn dialog and model-graded evaluations, plus 'a collection of over 200 pre-built evaluations ready to run on any model'.