SoftwareFramework

Inspect-Hawk (Hawk)

Project record

Maintained by METR (Model Evaluation & Threat Research)1 2

Inspect-Hawk is METR's open-source platform for running Inspect AI evaluations on cloud infrastructure. Users define tasks, agents, and models in a YAML file; Hawk runs each evaluation in an isolated Kubernetes pod, manages model API credentials through a proxy, streams logs, stores results in a PostgreSQL warehouse, and serves a web interface for browsing them. It is built on Inspect AI, the evaluation framework created by the UK AI Security Institute.1 3Fact reviewed Oct 2, 2026

Repository: Repository (external site: github.com)
Documentation: Documentation (external site: hawk.metr.org)
License: LICENSE (MIT) (external site: github.com)
Release notes: Releases (external site: github.com)

Last reviewedOct 2, 2026Entry updated Oct 2, 2026Documented release Unknown

Availability and license

Overall availability

Public

Source code is public on GitHub and the CLI is installed from PyPI. Running evaluations at scale requires deploying Hawk to the user's own AWS account (EKS, ECS Fargate, Aurora PostgreSQL, S3, and related services) and supplying model provider access.1 3

Availability fact review: Oct 2, 2026

Availability is separate from permission: read the license before using or redistributing.

MIT License (external site: github.com)2
Applies to: Code · SPDX MIT · on the rubric’s OSI-approved software-license list · fact reviewed Oct 2, 2026

The MIT license covers Hawk's own code. It does not cover the evaluated models or the cloud services Hawk deploys to, which are used under their providers' terms.2 1Fact reviewed Oct 2, 2026

Component reuse rights

A readable or downloadable component is not automatically reusable. These indicators concern recorded license evidence, not system certification.

weights: Unknown — no complete fact-level rights review
code: Reviewed qualifying license recorded — check scope and conditions
data: Unknown — no complete fact-level rights review
documentation: Unknown — no complete fact-level rights review

No complete system-rights review is recorded for this release.

Public materials checklist

Items for a framework under USASI rubric v0.2. Unknown means unassessed or insufficient evidence.

Public materials checklist for Inspect-Hawk (Hawk)
Item	Status	Notes and evidence
Source codeIs the source code publicly readable?	Public	Public repository in the METR GitHub organization.1
DocumentationIs user documentation published?	Public	Documentation site at hawk.metr.org with deployment, configuration, CLI, and evaluation guides.3 1
InstallationAre installation instructions or packages publicly available?	Public	The README installs the CLI from PyPI with uv or pip (package hawk with the cli extra) and describes deploying an instance with Pulumi, Docker, Python, Node.js, and uv, plus a domain name.1
Supported platformsAre supported operating systems or hardware documented?	Public	Deployment targets AWS; a local mode runs the same evaluation configuration on the user's machine for debugging. The built-in model proxy covers OpenAI, Anthropic, and Google Vertex.1 3
Release statusAre versioned releases published?	Public	Versioned GitHub releases are published; the latest at review was v3.6.0 (September 24, 2026).4

What it is useful for

Teams that run model evaluations regularly and want to run grids of tasks, agents, and models on their own AWS deployment, control which users can run or view results for each model, and run Inspect Scout scans over transcripts from earlier runs.1Fact reviewed Oct 2, 2026

Run and use notes

Documented facts only. No hardware or performance claims are made without a cited source and stated assumptions.

The README advises keeping the CLI version in sync with the deployment and documents a version check that warns before each command by default.1

Organization context

METR (Model Evaluation & Threat Research)

Provenance and derivatives

Hawk is an original METR repository, not a fork. It depends on Inspect AI, the open-source evaluation framework created by the UK AI Security Institute, and can run scans with Inspect Scout, which is maintained in the meridianlabs-ai GitHub organization.1

Derived from: Inspect AI (UK AI Security Institute) (external site: inspect.aisi.org.uk) — Evaluation framework Hawk runs; not a catalog member.

U.S. eligibility

Project eligibility rests on documented governing or maintaining entities, not on contributors.

Eligible · basis: U.S.-governed project

Hawk is published in METR's GitHub organization, and its MIT license names METR as the 2026 copyright holder. METR (Model Evaluation and Threat Research, Inc.) is listed in the IRS exempt-organization extract for California as a Section 501(c)(3) organization, and its staff work in Berkeley, California (see the METR record). Hawk builds on Inspect AI, which the README attributes to the UK AI Security Institute; that dependency is recorded as provenance and is not a catalog member.1 2 5

Assessed Oct 2, 2026

Sources

1.
METR/hawk (GitHub repository and README) (external site: github.com)
METR (GitHub) · Repository · accessed Oct 2, 2026 · evidence reviewed Oct 2, 2026
2.
METR/hawk LICENSE (MIT) (external site: raw.githubusercontent.com)
METR (GitHub) · License · accessed Oct 2, 2026 · evidence reviewed Oct 2, 2026
3.
Inspect Hawk documentation (external site: hawk.metr.org)
METR · Documentation · accessed Oct 2, 2026 · evidence reviewed Oct 2, 2026
4.
METR/hawk releases (external site: github.com)
METR (GitHub) · Release notes · accessed Oct 2, 2026 · evidence reviewed Oct 2, 2026
5.
Exempt Organizations Business Master File Extract - California (eo_ca.csv) (external site: irs.gov)
Internal Revenue Service · Filing · accessed Oct 2, 2026 · evidence reviewed Oct 2, 2026

Email a correction Report a correction Edit this entry

This listing is not an endorsement, a safety assessment, or a federal approval.