# Leaderboard Data Directory

> Where the raw data behind 40 LLM benchmark leaderboards lives (GitHub, Hugging Face, API, CSV), what its licence or terms allow (quoted verbatim with the source URL), whether outside submissions are open, and the newest update date read from the source. Checked 2026-10-31. No benchmark scores are republished. "not published" means we found nothing on the pages we checked. Not legal advice.

Machine-readable data: https://benchdata.branolabs.com/api/leaderboards.json
Per-leaderboard pages: https://benchdata.branolabs.com/l/<slug>
Changes: https://benchdata.branolabs.com/changelog (RSS https://benchdata.branolabs.com/feed.xml)

## Leaderboards
- Aider LLM Leaderboards (polyglot) (Aider (Paul Gauthier)): raw data GitHub file https://github.com/Aider-AI/aider/blob/main/aider/website/_data/polyglot_leaderboard.yml; licence Apache-2.0 (Reuse with attribution); submissions Not stated; last updated 2025-11-20. https://benchdata.branolabs.com/l/aider-polyglot
- AlpacaEval (Stanford Tatsu Lab): raw data GitHub repo https://github.com/tatsu-lab/alpaca_eval; licence Apache-2.0 (repo) (Reuse with attribution); submissions Open; last updated 2025-08-09. https://benchdata.branolabs.com/l/alpacaeval
- ARC Prize leaderboard (ARC-AGI) (ARC Prize Foundation): raw data not published; licence No licence found for results (No licence found); submissions Restricted; last updated not published. https://benchdata.branolabs.com/l/arc-prize
- Artificial Analysis (Artificial Analysis): raw data API https://artificialanalysis.ai/data-api/docs; licence Proprietary terms (Reuse restricted by terms); submissions Not stated; last updated not published. https://benchdata.branolabs.com/l/artificial-analysis
- BenchLeader (BenchLeader): raw data JSON / CSV https://www.benchleader.com/data/leaderboard.json; licence CC BY 4.0 (own figures) (Reuse with attribution); submissions Not stated; last updated not published. https://benchdata.branolabs.com/l/benchleader
- BenchLM (BenchLM.ai): raw data JSON download / API https://benchlm.ai/data; licence CC BY-NC 4.0 (Non-commercial only); submissions Not stated; last updated not published. https://benchdata.branolabs.com/l/benchlm
- Berkeley Function Calling Leaderboard (BFCL) (UC Berkeley (Gorilla)): raw data GitHub repo https://github.com/ShishirPatil/gorilla/tree/main/berkeley-function-call-leaderboard; licence Apache-2.0 (Unclear); submissions Not stated; last updated 2026-04-12. https://benchdata.branolabs.com/l/bfcl
- BigCodeBench (BigCode project): raw data HF space https://huggingface.co/spaces/bigcode/bigcodebench-leaderboard; licence Apache-2.0 (code) (Unclear); submissions Closed; last updated 2025-10-15. https://benchdata.branolabs.com/l/bigcodebench
- Epoch AI Benchmarking Hub (Epoch AI): raw data ZIP download https://epoch.ai/benchmarks/use-this-data; licence CC BY 4.0 (Reuse with attribution); submissions Not stated; last updated 2026-10-31. https://benchdata.branolabs.com/l/epoch-ai-benchmarking-hub
- EQ-Bench (Samuel Paech): raw data not published; licence MIT (Unclear); submissions Restricted; last updated 2026-05-31. https://benchdata.branolabs.com/l/eq-bench
- EvalPlus (HumanEval+ / MBPP+) (EvalPlus team): raw data not published; licence Apache-2.0 (code) (Unclear); submissions Not stated; last updated 2025-10-02. https://benchdata.branolabs.com/l/evalplus
- GAIA (GAIA team (Meta, Hugging Face)): raw data HF dataset https://huggingface.co/datasets/gaia-benchmark/results_public; licence No licence on the dataset (No licence found); submissions Not stated; last updated 2026-10-31. https://benchdata.branolabs.com/l/gaia
- HELM (Stanford CRFM): raw data Public GCS bucket https://crfm-helm.readthedocs.io/en/latest/downloading_raw_results/; licence Apache-2.0 (Unclear); submissions Not stated; last updated 2026-06-05. https://benchdata.branolabs.com/l/helm
- Humanity's Last Exam (HLE) (Center for AI Safety and Scale AI): raw data not published; licence MIT (code) (Unclear); submissions Not stated; last updated 2026-09-22. https://benchdata.branolabs.com/l/humanitys-last-exam
- K2 Vendor Verifier (Moonshot AI): raw data GitHub repo https://github.com/MoonshotAI/K2-Vendor-Verifier; licence No licence (No licence found); submissions Not stated; last updated 2026-02-14. https://benchdata.branolabs.com/l/k2-vendor-verifier
- Kagi LLM Benchmarking Project (Kagi Inc.): raw data not published; licence No licence found (No licence found); submissions Not stated; last updated 2026-02-13. https://benchdata.branolabs.com/l/kagi-llm-benchmark
- LiveBench (LiveBench team / Abacus AI): raw data HF dataset https://huggingface.co/datasets/livebench/model_judgment; HF dataset https://huggingface.co/datasets/livebench/model_answer; licence Apache-2.0 (copied from FastChat) (Unclear); submissions Not stated; last updated 2026-10-25. https://benchdata.branolabs.com/l/livebench
- LiveCodeBench (LiveCodeBench team): raw data not published; licence MIT (Unclear); submissions Not stated; last updated 2025-07-16. https://benchdata.branolabs.com/l/livecodebench
- LLM Confabulation Leaderboard (Lech Mazur): raw data GitHub repo https://github.com/lechmazur/confabulations; licence No licence (No licence found); submissions Not stated; last updated 2025-08-07. https://benchdata.branolabs.com/l/confabulations
- LLM Stats (llm-stats.com): raw data not published; licence Reuse with attribution (site terms) (Reuse with attribution); submissions Not stated; last updated not published. https://benchdata.branolabs.com/l/llm-stats
- LMArena (Arena) leaderboard (LMArena): raw data HF dataset https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset; licence CC BY 4.0 (Reuse with attribution); submissions Not stated; last updated 2026-10-30. https://benchdata.branolabs.com/l/lmarena
- MathArena (SRI Lab, ETH Zurich): raw data HF datasets https://huggingface.co/MathArena; licence MIT (Unclear); submissions Not stated; last updated 2026-10-19. https://benchdata.branolabs.com/l/matharena
- MMMU / MMMU-Pro (MMMU team): raw data not published; licence Site template CC BY-SA 4.0 (Unclear); submissions Not stated; last updated 2025-09-05. https://benchdata.branolabs.com/l/mmmu
- MTEB Leaderboard (MTEB (embeddings-benchmark)): raw data GitHub repo https://github.com/embeddings-benchmark/results; HF dataset https://huggingface.co/datasets/mteb/results; licence CC0 1.0 (Public domain (CC0)); submissions Open; last updated 2026-10-30. https://benchdata.branolabs.com/l/mteb
- NYT Connections (Extended) (Lech Mazur): raw data GitHub repo https://github.com/lechmazur/nyt-connections; licence No licence (No licence found); submissions Not stated; last updated 2026-10-23. https://benchdata.branolabs.com/l/nyt-connections
- Open LLM Leaderboard (Hugging Face): raw data HF dataset https://huggingface.co/datasets/open-llm-leaderboard/contents; licence No licence on the dataset (No licence found); submissions Closed; last updated 2025-03-20. https://benchdata.branolabs.com/l/open-llm-leaderboard
- OpenRouter rankings (OpenRouter, Inc.): raw data not published; licence No licence found (No licence found); submissions Not stated; last updated 2026-10-30. https://benchdata.branolabs.com/l/openrouter-rankings
- OpenVLM Leaderboard (OpenCompass) (OpenCompass (Shanghai AI Lab)): raw data JSON download http://opencompass.openxlab.space/assets/OpenVLM.json; licence Apache-2.0 (Unclear); submissions Not stated; last updated 2026-03-11. https://benchdata.branolabs.com/l/open-vlm-leaderboard
- OSWorld / OSWorld-Verified (XLANG Lab (HKU)): raw data not published; licence Apache-2.0 (Unclear); submissions Not stated; last updated 2026-10-24. https://benchdata.branolabs.com/l/osworld
- QuantDiff (QuantDiff): raw data CSV / JSON export https://quantdiff.com/data/export/README.md; licence CC BY 4.0 (Reuse with attribution); submissions Not stated; last updated 2026-10-30. https://benchdata.branolabs.com/l/quantdiff
- RewardBench / RewardBench 2 (Allen Institute for AI (Ai2)): raw data HF dataset https://huggingface.co/datasets/allenai/reward-bench-2-results; HF dataset https://huggingface.co/datasets/allenai/reward-bench-results; licence Apache-2.0 (Unclear); submissions Not stated; last updated 2025-12-11. https://benchdata.branolabs.com/l/rewardbench
- Scale SEAL Leaderboards (Scale AI): raw data not published; licence No licence found (No licence found); submissions Restricted; last updated not published. https://benchdata.branolabs.com/l/scale-seal
- SimpleBench (SimpleBench Team): raw data not published; licence MIT (code) (Unclear); submissions Not stated; last updated 2024-12-20. https://benchdata.branolabs.com/l/simplebench
- SWE-bench (SWE-bench team (Princeton / Stanford)): raw data GitHub repo https://github.com/SWE-bench/experiments; licence No licence on the experiments repo (No licence found); submissions Restricted; last updated 2026-09-01. https://benchdata.branolabs.com/l/swe-bench
- Terminal-Bench (Stanford / Laude Institute (Harbor)): raw data HF dataset https://huggingface.co/datasets/harborframework/terminal-bench-2-leaderboard; licence Apache-2.0 (Reuse with attribution); submissions Closed; last updated 2026-05-15. https://benchdata.branolabs.com/l/terminal-bench
- UGI Leaderboard (DontPlanToEnd): raw data not published; licence Apache-2.0 (Space) (Unclear); submissions Not stated; last updated 2026-10-17. https://benchdata.branolabs.com/l/ugi-leaderboard
- Vals AI benchmarks (Vals AI): raw data not published; licence No licence found (No licence found); submissions Not stated; last updated 2026-10-29. https://benchdata.branolabs.com/l/vals-ai
- Vectara Hallucination Leaderboard (Vectara): raw data GitHub repo https://github.com/vectara/hallucination-leaderboard; licence Apache-2.0 (Reuse with attribution); submissions Not stated; last updated 2026-10-23. https://benchdata.branolabs.com/l/vectara-hallucination
- WildBench (Allen Institute for AI (Ai2)): raw data not published; licence Apache-2.0 (code) (Unclear); submissions Not stated; last updated 2024-08-01. https://benchdata.branolabs.com/l/wildbench
- τ-bench (tau2-bench) (Sierra): raw data GitHub repo https://github.com/sierra-research/tau2-bench; licence MIT (repo) (Reuse with attribution); submissions Open; last updated 2026-10-30. https://benchdata.branolabs.com/l/tau-bench

## Pages
- [Methodology](https://benchdata.branolabs.com/about)
- [Changelog](https://benchdata.branolabs.com/changelog)

## Optional
- [@BranoLabs on X](https://x.com/BranoLabs)
- [BranoLabs](https://branolabs.com): more small, data-backed AI tools