HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen3-32B

Qwen/Qwen3-32B

family qwen · 26 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen3-32B
best scores
GPQA: 68.4MMLU: 90.9AIME 2024: 81.4MATH: 92.0DeepSWE: 42.2MT-Bench: 9.5WildBench: 67.57

Best scores per benchmark

benchmarkcategorybest score
MATHmath92.0
MMLUknowledge90.9
AIME 2024math81.4
GPQAreasoning68.4
WildBenchchat67.57
DeepSWEswe42.2
MT-Benchchat9.5

Variants (26)

variantkindquantbitsvRAMdlorgyear
Qwen3-32Boriginalbf1616.010,114,576Qwen2025
Qwen3-32B-AWQawqawq2,005,871Qwen2025
Qwen3-32B-FP8fp8fp88.079,647Qwen2025
Qwen3-32B-bnb-4bitbnb-4bit4bit76,451unsloth2025
Qwen3-32B-NVFP4nvfp4nvfp44.070,912nvidia2025
Qwen3-32B-GGUFggufgguf16.026,368unsloth2025
Qwen3-32B-GGUFggufgguf4.024,933Qwen2025
Qwen3-32B-MLX-4bitmlxmlx14,912lmstudio-community2025
Qwen3-32B-MLX-8bitmlxmlx14,477lmstudio-community2025
Qwen3-32B-GGUFggufgguf3.011,443lmstudio-community2025
Qwen3-32Boriginalbf1616.08,824unsloth2025
Qwen3-32B-unsloth-bnb-4bitbnb-4bit4bit6,753unsloth2025
Qwen_Qwen3-32B-GGUFggufgguf2.06,015bartowski2025
Qwen3-32B-128K-GGUFggufgguf16.01,898unsloth2025
Qwen3-32B-GGUFggufgguf4.01,646mradermacher2025
Qwen3-Nemotron-32B-GenRM-Principleoriginalbf1616.01,130nvidia2025
FrogBoss-32B-2510originalbf1616.0774microsoft2026
Qwen3-32B-MLX-8bitmlxmlx756Qwen2025
Qwen3-32B-MLX-4bitmlxmlx573Qwen2025
Qwen3-32B-MLX-bf16mlxmlx467Qwen2025
Qwen3-32B-GGUFggufgguf16.0276Mungert2025
Qwen3-32B-GGUFggufgguf4.0210ggml-org2025
Qwen3-32B-MLX-6bitmlxmlx192Qwen2025
Qwen3-Nemotron-32B-RLBFForiginalbf1616.0138nvidia2025
Qwen-3-Nemotron-32B-Rewardoriginalbf1616.068nvidia2025
Qwen3-32B-AWorldoriginalbf1616.032inclusionAI2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.