Purpose and responsibility
About AI Bench Index
Catalogue review recorded · individual claims have separate datesReferenceMethodology
A technology and AI information website
AI Bench Index explains AI models, evaluations, developer tools and local computing. Its purpose is to help readers understand the technology and the evidence: what a number measures, which conditions apply, what changed and what is still unknown.
It is not a decision engine or a personalised adviser. The sole standalone tool is the PC parts-and-prices builder. The catalogues use ordinary search, filters, tables and expandable explanations to make information easier to inspect.
What is included—and what is not
We focus on subjects for which the existing reference pages can offer useful depth: documented model specifications, dated commercial terms, published evaluation results, hardware constraints and reproducible measurements. Inclusion is not an endorsement. Catalogue coverage is incomplete, and a missing product or score is not evidence of poor quality.
Publisher reports, independently reported evaluations, our own measurements and calculated examples are different kinds of evidence. A source link does not mean that we reproduced a test. No product is described as hands-on tested without an actual recorded run.
Useful detail belongs beside the relevant entry. More database records do not automatically create more pages; new public pages require a separate editorial decision.
How this site is maintained
AI assists research, drafting and development. It is not an expert credential or a substitute for primary evidence. The site owner is responsible for publication approval. We do not imply a larger reporting team, laboratory capacity or testing history than the records support.
Price dates describe when a quote was checked. Source-review dates describe a recorded check, not a new experiment. Historical results remain historical; missing configuration fields stay unknown.
Read the evidence standard and correction recordRequesting a correction
Include the page URL, exact entry or claim, what appears wrong, and a primary source with a publication date. A useful correction explains whether the old value was wrong at the time or has since changed; those require different treatment.
Corrections are recorded with their reason and source, then checked across the pages that reuse that record. Valid old observations are retained rather than silently rewritten.
To request a correction, email support@aibenchindex.com with the page URL, the exact claim, the proposed correction and a dated primary source. AI Bench Index records material corrections and preserves valid historical observations.