Show the work behind the claim.

What is available today

The tool guide is based on official documentation. No original hands-on test results have been published as part of this release. Existing practical guides describe suggested workflows, not measured product performance.

What a hands-on test must record

• The task, success criteria and exact input material (or a reproducible, privacy-safe fixture).
• Product, underlying model/version, subscription, test date, tools, settings and prompts.
• Unedited outputs, errors, refusals and any interventions or retries.
• Time for generation, verification and editing; usage and costs where observable.
• A per-run result against each stated criterion, with evidence links.
• Number of repeated runs and what changed between them. A single observation remains a single observation.
• Known limitations, excluded tasks and the conditions under which the conclusion applies.

How to read our formats

News

What changed, attributed to identifiable sources.

Analysis

Our interpretation, separated from reported facts.

Documentation-based assessment

Published capabilities and restrictions; no claim of hands-on experience.

Practical guide

A goal, prerequisites, steps, examples, expected output and troubleshooting.

Hands-on test

Results backed by an actual run record and its limitations.

Review and corrections

Tool records are queued for review after 30 days, or sooner following a relevant product change. A date changes only when facts have been checked or content materially updated. A scheduled review is not evidence of a completed check.

Report a correction through our contact page, including the page URL, disputed claim and a supporting source. Material corrections are described in the affected page's review history.

Editorial responsibility rests with founder Nikolaj Lysgaard. Read the editorial policy.