The AI resource library

AI statistics, benchmarks & research.

Find the numbers behind the headlines, compare model performance, and understand how AI works. A curated collection of websites worth exploring.

190 information websites7 resource typesSources checked
Compare AI benchmark scoresBrowse AI tools

190 of 190 websites

  • Benchmarks & leaderboards

    Artificial Analysis

    Compare AI models and hosting providers by benchmark results, speed, latency and price.

    Useful for Shortlisting models before testing your own workload.

    Reading note

    Read the methodology and configuration behind each comparison.

    artificialanalysis.aiVisit site
  • Benchmarks & leaderboards

    Arena

    Explore model leaderboards based on human preferences across supported text and media tasks.

    Useful for Seeing how users rate competing model outputs.

    Reading note

    Preference rankings depend on the voter pool, task mix and uncertainty.

    arena.aiVisit site
  • Statistics & trends

    Epoch AI

    Explore research and datasets on AI capability, compute, training and broader industry trends.

    Useful for Understanding the evidence behind AI progress claims.

    Reading note

    Check the date, definitions and uncertainty of each dataset.

    epoch.aiVisit site
  • Statistics & trends

    Stanford AI Index

    Read annual reports and supporting data about AI research, investment, adoption and social impact.

    Useful for Finding statistics for business plans and presentations.

    Reading note

    Annual reports describe a dated snapshot, not live market totals.

    hai.stanford.eduVisit site
  • Statistics & trends

    Our World in Data AI

    Browse charts, data and explanations about the development and impact of artificial intelligence.

    Useful for Finding understandable, source-linked AI statistics.

    Reading note

    Read each chart's underlying source and measurement notes.

    ourworldindata.orgVisit site
  • Policy & responsible AI

    OECD AI Policy Observatory

    Explore AI policy information, country activity and data from the OECD's AI observatory.

    Useful for Comparing AI policy developments across countries.

    Reading note

    Policy trackers are reference material; confirm current rules with the relevant authority.

    oecd.aiVisit site
  • Benchmarks & leaderboards

    HELM

    Explore Stanford's framework and results for evaluating language models across several dimensions.

    Useful for Looking beyond a single overall model score.

    Reading note

    Compare models within the same evaluation setting and version.

    crfm.stanford.eduVisit site
  • Benchmarks & leaderboards

    MLCommons

    Find standardised machine-learning benchmarks and published results for systems and workloads.

    Useful for Comparing AI hardware and system performance.

    Reading note

    Hardware, software and benchmark division affect comparability.

    mlcommons.orgVisit site
  • Benchmarks & leaderboards

    SWE-bench

    View software-engineering benchmark results based on resolving issues in real code repositories.

    Useful for Assessing coding agents on repository-level tasks.

    Reading note

    Check the benchmark variant, agent setup and evaluation rules.

    swebench.comVisit site
  • Benchmarks & leaderboards

    Terminal-Bench

    Explore evaluations of agents completing tasks in a terminal environment.

    Useful for Understanding command-line and agent-workflow capabilities.

    Reading note

    Results depend on the agent, model, tools and benchmark release.

    tbench.aiVisit site
  • Benchmarks & leaderboards

    LiveBench

    Compare language-model results across periodically refreshed evaluation tasks and releases.

    Useful for Checking performance across several task categories.

    Reading note

    Use the same release when comparing scores.

    livebench.aiVisit site
  • Benchmarks & leaderboards

    LiveCodeBench

    Explore coding-model evaluations built from time-stamped programming problems.

    Useful for Comparing coding performance with attention to data contamination.

    Reading note

    Task dates and evaluation windows affect the reported result.

    livecodebench.github.ioVisit site
All 190 websites — quick index