HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen2.5-32B

Qwen/Qwen2.5-32B

family qwen · 13 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen2.5-32B
best scores
GPQA: 31.6MMLU-Pro: 47.5Humanity's Last Exam: 11.8LiveCodeBench: 57.2SciCode: 100.0AIME 2024: 83.3AIME 2025: 70.0BBH: 84.5GPQA Diamond: 63.1MATH: 48.9MATH-500: 88.4Math Olympiad: 55.0

Best scores per benchmark

benchmarkcategorybest score
SciCodecoding100.0
MATH-500math88.4
BBHreasoning84.5
AIME 2024math83.3
AIME 2025math70.0
GPQA Diamondreasoning63.1
LiveCodeBenchcoding57.2
Math Olympiadmath55.0
MATHmath48.9
MMLU-Proknowledge47.5
GPQAreasoning31.6
Humanity's Last Examknowledge11.8

Variants (13)

variantkindquantbitsvRAMdlorgyear
Qwen2.5-32B-Instructoriginalbf1616.01,396,492Qwen2024
QwQ-32Boriginalbf1616.0378,909Qwen2025
Qwen2.5-32Boriginalbf1616.049,521Qwen2024
OpenReasoning-Nemotron-32Boriginalbf1616.09,429nvidia2025
Qwen2.5-Coder-32Boriginalbf1616.01,699Qwen2024
Qwen2.5-32B-bnb-4bitbnb-4bit4bit610unsloth2025
QwQ-32B-GGUFggufgguf16.0269Mungert2025
LongWriter-Zero-32B-GGUFggufgguf16.0267Mungert2025
OpenMath-Nemotron-32Boriginalbf1616.0209nvidia2025
SynLogic-32Boriginalbf1616.0111MiniMaxAI2025
SynLogic-Mix-3-32Boriginalbf1616.0110MiniMaxAI2025
NFT-32Boriginalbf1616.072nvidia2025
OpenMath-Nemotron-32B-GGUFggufgguf16.060Mungert2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.