How LITHOVEX AI model data is maintained

How the catalog is assembled

The generation pipeline reads model choices from the application settings and merges them with the maintained SEO catalog records so existing catalog-only details are retained. Public model profiles render those records; the generation process does not fetch or verify model facts from provider APIs.

The pipeline creates model-pair records from the catalog. The comparison pages display recorded values and calculate only derivations supported by those values, such as shared benchmark differences, numeric input/output token-cost examples, and context window ratios. If the necessary values are absent or not numeric, the page says the comparison cannot be calculated.

Comparison eligibility

A pair is indexable only when at least one catalog record lists the other as an alternative or competitor, their inferred modality and task family are compatible, and they have either two or more shared populated benchmark labels or both models have pricing and context window data. Other generated pair records remain noindex and are not linked as indexed comparisons. This is a catalog eligibility rule, not a claim that the models were tested head-to-head.

Freshness, limitations, and corrections

The catalog does not preserve a source URL and verified-at timestamp for every field, or a trustworthy per-model last-verified date. A build date is not treated as a data update. Benchmark entries may use different versions and evaluation setups; they are reference data, not controlled comparative results. Check provider documentation before making a purchase or deployment decision.

To report an error or request a correction, contact LITHOVEX AI support at lithovex@gmail.com.

Data last updated: the catalog does not record a reliable update date.