Knowledge Bases and Evaluation Organizations
Knowledge Bases and Evaluation Organizations (Knowledge bases, benchmarks and public evaluations) holds 117 entries from 15 sources. Larger sources have their own directory pages, 15 entries per page. Authority tiers in this category: A2–B. Admission policy: All core resources are included. Each secondary source is admitted according to its own database, methodology, versions and leaderboard structure; there is no cross-organization scoring scheme.
Highlights by source
- MITRE ATLAS (4 entries): MITRE ATLAS: knowledge base of adversarial threats to AI; MITRE ATLAS Fact Sheet; MITRE ATLAS Data and more
- OWASP GenAI Security Project (8 entries): OWASP GenAI LLM Top 10 2026; OWASP Top 10 for Agentic Applications 2026; Agent Control Standard（ACS） and more
- AI Incident Database (5 entries): AI Incident Database; AI Incident Roundup: May–July 2026; AIID Taxonomies and more
- MIT AI Risk Repository (7 entries): MIT AI Risk Repository; MIT AI Risk Mitigation Map; MIT AI Risk Initiative data and research hub and more
- MLCommons (14 entries): MLCommons Benchmark Suites; MLPerf Training; MLPerf Inference and more
- METR (8 entries): Measuring AI Ability to Complete Long Software Tasks; Task-Completion Time Horizons of Frontier AI Models; Frontier Risk Report (February–March 2026) and more
- ARC Prize Foundation (8 entries): ARC-AGI-3: an interactive reasoning benchmark; ARC-AGI-3 Technical Report; ARC-AGI Official Leaderboard and more
- Stanford CRFM / HELM (12 entries): HELM：Holistic Evaluation of Language Models; 2025 Foundation Model Transparency Index; HELM Capabilities and more
- LMArena (formerly LMSYS / Arena) (10 entries): Arena：How It Works; Arena Leaderboard Policy; Agent Arena：Causal Evaluation of Agents in the Real World and more
- Artificial Analysis (12 entries): Artificial Analysis Models Comparison; Artificial Analysis Intelligence Benchmarking Methodology; Artificial Analysis API Performance Benchmarking Methodology and more
- OpenCompass (Sinan) (6 entries): OpenCompass public academic leaderboard for large models; Guide to reproducing CompassAcademic leaderboard results; OpenCompass evaluation framework overview and more
- SuperCLUE (4 entries): SuperCLUE general leaderboard; SuperCLUE large model evaluation system; SuperCLUE Agent Benchmark 2025H1 and more
- FlagEval (BAAI) (5 entries): FlagEval (Tianping) platform user manual and evaluation portal; FlagEval (Tianping) evaluation system; FlagEval evaluation workflow and more
- SWE-bench (7 entries): SWE-bench Official Leaderboards; SWE-bench: a benchmark for resolving real GitHub issues; SWE-bench Dataset Catalog and more
- Berkeley BFCL (7 entries): Berkeley Function Calling Leaderboard V4; Berkeley Function-Calling Leaderboard Methodology; BFCL V4：Agentic Evaluation and more