HF Model & Benchmark Matrix

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

Qwen/Qwen2.5-VL-32B-Instruct

family qwen · 9 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen2.5-VL-32B-Instruct
best scores
MMLU: 78.4MMLU-Pro: 68.8MATH: 82.2GPQA Diamond: 46.0MBPP: 84.0MMMU: 70.2MathVista: 74.8DocVQA: 96.4MMBench: 87.3OSWorld: 8.83

Best scores per benchmark

benchmarkcategorybest score
DocVQAmultimodal96.4
MMBenchmultimodal87.3
MBPPcoding84.0
MATHmath82.2
MMLUknowledge78.4
MathVistamultimodal74.8
MMMUmultimodal70.2
MMLU-Proknowledge68.8
GPQA Diamondreasoning46.0
OSWorldagentic8.83

Variants (9)

variantkindquantbitsvRAMdlorgyear
Qwen2.5-VL-32B-Instructoriginalbf1616.0777,903Qwen2025
Qwen2.5-VL-32B-Instruct-AWQawqawq762,238Qwen2025
Qwen2.5-VL-32B-Instruct-GGUFggufgguf3.07,903lmstudio-community2025
Qwen_Qwen2.5-VL-32B-Instruct-GGUFggufgguf2.06,002bartowski2025
Qwen2.5-VL-32B-Instruct-GGUFggufgguf4.04,150ggml-org2025
Qwen2.5-VL-32B-Instruct-GGUFggufgguf16.02,829unsloth2025
Qwen2.5-VL-32B-Instruct-bnb-4bitbnb-4bit4bit972unsloth2025
Qwen2.5-VL-32B-Instruct-GGUFggufgguf16.0731Mungert2025
Qwen2.5-VL-32B-Instruct-unsloth-bnb-4bitbnb-4bit4bit358unsloth2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.