Modelrovia
Our editorial standards.
How we research, test, and explain AI products—and make the limits of our evidence clear.
Research and testing are different
A guide may explain a workflow or summarize official documentation. A hands-on review requires actual use, retained evidence, a test date, and the tested product or model version. Unverified drafts stay unpublished.
Compare like with like
For comparisons, use the same input, task, and success criteria wherever possible. Record relevant settings, subscription tiers, retries, time, and costs. Show failures as well as successes.
Check access and pricing
Distinguish mainland and international products, web apps and APIs, and official services and third-party hosts. State which regions, payment methods, and English-language features were verified. Pricing needs a source and a check date.
Trace the evidence
Prefer official documentation for product facts. Cite policy and license pages for claims about commercial use or data handling. Clearly label interpretation and unresolved questions.
Use AI responsibly in the editorial process
AI can assist with research organization, drafting, and editing. It cannot stand in for a product test, create evidence of use, or justify an unsupported recommendation. Published content remains subject to editorial review.
Update with purpose
Change the update date when information is substantively checked or revised. Preserve stable URLs and redirect changed paths. Disclose commercial relationships alongside relevant content.