# RankedAGI

RankedAGI is a static public website for comparing frontier AI models by benchmark performance. It collects model results from official provider releases, benchmark leaderboards, papers, and public announcements, then presents them as ranked tables and model detail pages.

## Public routes

- Homepage: https://rankedagi.com/
- All models: https://rankedagi.com/models
- Sources and methodology: https://rankedagi.com/sources
- The RankedAGI Engine (scoring and simulated data): https://rankedagi.com/engine
- Sitemap: https://rankedagi.com/sitemap.xml

## Data scope

RankedAGI tracks public AI model benchmark scores, model metadata, context-window values, release/update dates, provider names, and composite RankedAGI scores. Benchmarks include coding, reasoning, math, general preference, multimodal, and agentic evaluations where public results are available.

## Methodology

Data sourcing, benchmark inclusion, update frequency, model versioning, corrections, and limitations:

https://rankedagi.com/sources

The composite RankedAGI benchmarks (Coding, Agentic, Reasoning, Math, Overall) and the simulated benchmark data that powers them — how missing values are estimated, how estimates feed composite scores, confidence handling, and how accuracy is validated:

https://rankedagi.com/engine

## Access

The production site is static and read-only. Use the public pages and sitemap for discovery.