{
    "GRIND": {
        "source": "Vellum GRIND (request access)",
        "url": "https://www.vellum.ai/llm-leaderboard",
        "notes": "Run via Vellum’s hosted harness; mirror prompts and scoring locally where possible.",
        "checksum": null
    },
    "AIME_2024": {
        "url": "https://huggingface.co/datasets/math-ai/aime24",
        "license": "see dataset card",
        "checksum": null
    },
    "GPQA": {
        "url": "https://huggingface.co/datasets/Idavidrein/gpqa",
        "paper": "https://arxiv.org/abs/2311.12022",
        "license": "MIT (dataset repo)",
        "checksum": null
    },
    "SWE_bench": {
        "url": "https://github.com/SWE-bench/SWE-bench",
        "leaderboard": "https://www.swebench.com/",
        "verified_subset": "https://openai.com/index/introducing-swe-bench-verified/",
        "checksum": null
    },
    "MATH_500": {
        "url": "https://huggingface.co/datasets/HuggingFaceH4/MATH-500",
        "origin": "subset of Hendrycks MATH (per dataset card)",
        "checksum": null
    },
    "BFCL": {
        "url": "https://gorilla.cs.berkeley.edu/leaderboard.html",
        "blog": "https://gorilla.cs.berkeley.edu/blogs/8_berkeley_function_calling_leaderboard.html",
        "checksum": null
    },
    "Aider_Polyglot": {
        "url": "https://github.com/Aider-AI/polyglot-benchmark",
        "leaderboard": "https://aider.chat/docs/leaderboards/",
        "notes": "225 Exercism problems across 6 languages",
        "checksum": null
    }
}