Use and attribution
AI Benchmark Radar is a research aid. These notes explain the scope of the information and how to approach reuse of content and linked resources.
Interpreting the information
Reported scores can depend on dataset versions, prompts, tools, budgets, and evaluation harnesses. A high score on one task does not establish broad superiority, reliability, or suitability for a particular deployment.
The directory may contain omissions, source errors, or outdated observations. Check the linked primary sources before relying on a result. Research summaries are not medical, financial, legal, or operational security advice.
Rights differ by material
Follow the license attached to the material you use. The upstream MIT software license covers software; it does not make papers, datasets, catalog descriptions, or editorial text MIT licensed.
The upstream technical report and original editorial content are identified under CC BY-NC-SA 4.0 with a separate commercial-permission requirement in the upstream notice. Third-party sources keep their own terms. Newly prepared website explanations are distinguished from preserved upstream descriptions.
Citations and links identify sources; they do not grant rights that the original owner has not granted. Reuse should preserve appropriate attribution and the context needed to avoid misleading readers.
External resources
External papers, software, datasets, and services are maintained by their respective providers. A directory link is not a guarantee of their availability, safety, licensing, or accuracy.
Cybersecurity evaluations should be run only in environments for which you have authorization. A research reference on this website does not grant permission to test a third-party system.