HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/meta-llama/meta-llama/Llama-3.1-70B

meta-llama/Llama-3.1-70B

family llama · 7 variants · unknown

provider
meta-llama
released
—
huggingface
https://huggingface.co/meta-llama/Llama-3.1-70B
best scores
MATH: 63.4MMLU: 84.4PopQA: 55.7TruthfulQA: 46.8BBH: 88.0DROP: 61.3GSM8K: 84.5HumanEval: 86.2IFEval: 80.51AlpacaEval: 12.4GPQA: 12.64MUSR: 24.49MMLU-Pro: 40.27

Best scores per benchmark

benchmarkcategorybest score
BBHreasoning88.0
HumanEvalcoding86.2
GSM8Kmath84.5
MMLUknowledge84.4
IFEvalinstruction-following80.51
MATHmath63.4
DROPreasoning61.3
PopQAknowledge55.7
TruthfulQAsafety46.8
MMLU-Proknowledge40.27
MUSRreasoning24.49
GPQAreasoning12.64
AlpacaEvalchat12.4

Variants (7)

variantkindquantbitsvRAMdlorgyear
Llama-3.3-70B-Instructoriginalbf1616.0635,792meta-llama2024
Llama-3.1-70Boriginalbf1616.043,290meta-llama2024
Meta-Llama-3.1-70B-bnb-4bitbnb-4bit4bit8,155unsloth2024
Llama-3.1-Tulu-3-70B-SFToriginalbf1616.0368allenai2025
cogito-v1-preview-llama-8B-GGUFggufgguf16.0294Mungert2025
Llama-3.3-70B-Instruct-GGUFggufgguf16.0291Mungert2025
OpenMath2-Llama3.1-70Boriginalbf1616.086nvidia2024
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.