Scores
Scores combine public provider claims, documentation, leaderboard-style signals, practical engineering fit, and editorial judgment. They should be used for shortlisting, not as a substitute for your own evals.
Trust
AI Model Rank Index is an editorial research tool, not an official benchmark from OpenAI, Anthropic, Google, xAI, Meta, DeepSeek, Mistral, Cohere, Perplexity, Runway, Pika, or Luma.
Scores combine public provider claims, documentation, leaderboard-style signals, practical engineering fit, and editorial judgment. They should be used for shortlisting, not as a substitute for your own evals.
API prices change often. Always verify token, image, video, caching, batch, and enterprise pricing on the provider's official pricing page before purchasing.
Local model recommendations are planning estimates. Real performance depends on quantization, runtime, drivers, cooling, operating system, context length, and background apps.