HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/meta-llama/meta-llama/Llama-3.3-70B-Instruct

meta-llama/Llama-3.3-70B-Instruct

family llama · 16 variants · unknown

provider
meta-llama
released
—
huggingface
https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct
best scores
MATH: 82.4

Best scores per benchmark

benchmarkcategorybest score
MATHmath82.4

Variants (16)

variantkindquantbitsvRAMdlorgyear
Llama-3.3-70B-Instruct-FP8fp8fp88.095,548nvidia2025
Llama-3.3-70B-Instruct-NVFP4nvfp4nvfp44.081,769nvidia2025
Llama-3.3-70B-Instructoriginalbf1616.027,723unsloth2025
Llama-3.3-70B-Instruct-GGUFggufgguf3.021,409lmstudio-community2024
Llama-3.3-70B-Instruct-GGUFggufgguf16.020,754unsloth2025
Llama-3.3-70B-Instruct-GGUFggufgguf1.015,399bartowski2024
Llama-3.3-70B-Instruct-bnb-4bitbnb-4bit4bit10,448unsloth2025
Llama-3.3-70B-Instruct-unsloth-bnb-4bitbnb-4bit4bit1,010unsloth2025
Llama-3.3-Nemotron-70B-Reward-Principleoriginalbf1616.0278nvidia2025
watt-tool-70B-GGUFggufgguf16.0249Mungert2025
Llama-3.3-Nemotron-70B-Reward-Multilingualoriginalbf1616.0195nvidia2025
Llama-3.3-70B-Instruct-FP8-Dynamicfp8fp88.0154unsloth2025
Llama-3.3-Nemotron-70B-Rewardoriginalbf1616.0101nvidia2025
Llama-3.3-Nemotron-70B-Selectoriginalbf1616.097nvidia2025
Llama-3.3-Nemotron-70B-Feedbackoriginalbf1616.064nvidia2025
Llama-3.3-Nemotron-70B-Editoriginalbf1616.055nvidia2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.