Model selection and API pricing

Find models by type, then validate on your own task.

Find models by type

Text model filters

Text evaluation focusMulti-select

Ranking priority

QualityMedium

Higher public-preference score in the selected focus

CostVery high

Lower disclosed reference cost

RequirementsOnly declared model facts; unknown never satisfies

Providers (optional) · All

    Checking scores, model facts and current quotes…

    More comparison tools

    How we compare

    Quality scores are Arena user preferences, not measurements on your task. We filter and match models; data provided by Arena under CC BY 4.0 · Arena Leaderboard Dataset ↗

    How we judge value

    Meet the task’s usability standard first, then compare cost per usable output, including failed calls and retries. Quotes help shortlist; the cheapest estimate does not prove the best result. Tasks, currencies and billing tiers stay separate. Insufficient evidence never becomes a best-model claim.

    Every price keeps its source and currency

    Provider endpoints keep paired input and output rates, route tags and quantization. Direct quotes keep the currency and region their provider bills in. Snapshots record when we observed a price change, not when a provider made it effective.

    What is not covered yet

    Coverage beyond the listed sources, context-tier and regional eligibility, categories without verified billing rates, free-tier allowances and our own performance measurements. Unknown values are shown as unknown and never counted as zero.

    Model coverage & sourcesLook up a specific modelProvider coveragePrice history & source status