Find the right AI for you

Search by name or tell us what you want to do
API servicesChoose how to access modelsAI modelsCapabilities, specifications, and pricingAI toolsFind useful tools by taskMCP directoryConnect tools, data, and workflowsAgent skillsFind reusable skills for AI agentsBenchmarksCompare model performanceOnline toolsConvert accounts, format JSON, estimate costs and test APIsModel timelineExplore model releasesUse casesStart with a practical task
EnterSearch · ↑↓Choose a resultEscClose
Submit中文
METHODOLOGY

Data & ranking rules

Understand the evidence behind a listing, what a comparison measures and where its conclusions apply.

Types of evidence

We distinguish official and provider claims, third-party benchmarks, our own measurements and user reports. Published specifications and observed behavior are recorded separately.

  • Evidence includes a reference, observation date, product version and relevant sampling conditions.
  • Missing information is distinguished from a service outage.
  • Technical rankings compare the stated benchmark scope, metrics and samples.

Price, reliability and speed

Price comparisons must match the model, access provider, currency, billing unit and cache conditions. Success rates need a time window, successful request count and total request count. Time to first token, generation speed and total duration are separate measurements.

  • Missing values are not zero.
  • A reachable website does not establish that a model API works.
  • Different test conditions are not joined into a continuous trend or a single overall score.

Service counts and model coverage

Counts cover our directory. A newly added listing is not necessarily a new business, and closure requires evidence. Access errors and unverified status are kept separate. Reopenings, identity merges and category changes are recorded.

Model coverage counts services listing a family. These counts can overlap; they are not traffic shares. Each service has one primary service type.

How rankings work

Price, reliability, speed, coverage and compatibility are assessed separately. Published technical rankings must state their comparison criteria, time window, minimum sample and rule version. We do not assign a technical rank without enough comparable evidence.

Advertising does not affect technical rankings. Editorial picks have separate reasons and update dates.

Corrections and revisions

We review public evidence before changing a listing. Historical changes retain their references and observation dates. Results recalculated under today’s rules are not presented as rankings published in the past.