HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen2.5-7B-Instruct

Qwen/Qwen2.5-7B-Instruct

family qwen · 22 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen2.5-7B-Instruct
best scores
DocVQA: 92.8ChartQA: 85.7MMMU: 69.1MMBench: 83.1MathVista: 63.8LiveCodeBench: 65.6GPQA: 38.6MMLU-Pro: 47.5Humanity's Last Exam: 11.8SciCode: 2.2AIME 2024: 55.5AIME 2025: 45.6MATH-500: 82.8Codeforces: 14.9BrowseComp: 100.0PopQA: 51.8MuSiQue: 35.8

Best scores per benchmark

benchmarkcategorybest score
BrowseCompagentic100.0
DocVQAmultimodal92.8
ChartQAmultimodal85.7
MMBenchmultimodal83.1
MATH-500math82.8
MMMUmultimodal69.1
LiveCodeBenchcoding65.6
MathVistamultimodal63.8
AIME 2024math55.5
PopQAknowledge51.8
MMLU-Proknowledge47.5
AIME 2025math45.6
GPQAreasoning38.6
MuSiQuereasoning35.8
Codeforcescoding14.9
Humanity's Last Examknowledge11.8
SciCodecoding2.2

Variants (22)

variantkindquantbitsvRAMdlorgyear
Qwen2.5-7B-Instruct-AWQawqawq4,446,238Qwen2024
Qwen2.5-7B-Instruct-GPTQ-Int4gptqgptq331,440Qwen2024
Qwen2.5-7B-Instruct-bnb-4bitbnb-4bit4bit251,443unsloth2025
Eagle2.5-8Bmergemerge129,409nvidia2025
Qwen2.5-7B-Instruct-GGUFggufgguf2.0122,082bartowski2024
Qwen2.5-7B-Instruct-unsloth-bnb-4bitbnb-4bit4bit89,343unsloth2025
Qwen2.5-7B-Instruct-GGUFggufgguf81,948Qwen2024
Qwen2.5-7B-Instructoriginalbf1616.076,543unsloth2025
Qwen2.5-7B-Instruct-GGUFggufgguf3.023,515lmstudio-community2024
Qwen2.5-7B-Instruct-GPTQ-Int8gptqgptq3,197Qwen2024
VulnLLM-R-7B-GGUFggufgguf16.02,371Mungert2026
Qwen2.5-7B-Instruct-MLX-4bitmlxmlx2,245lmstudio-community2024
Qwen2.5-7B-Instruct-i1-GGUFggufgguf1.01,770mradermacher2024
Qwen2.5-7B-Instruct-MLX-8bitmlxmlx1,265lmstudio-community2024
OpenCodeReasoning-Nemotron-7Boriginalbf1616.0593nvidia2025
HARC-Qwen2.5-7B-Instructmergemerge569microsoft2026
OpenReasoning-Nemotron-7B-GGUFggufgguf16.0473Mungert2025
OpenThinker3-7B-GGUFggufgguf16.0353Mungert2025
pokee_research_7b-GGUFggufgguf16.0248Mungert2025
OpenCodeReasoning-Nemotron-7B-GGUFggufgguf16.0110Mungert2025
OpenCodeReasoning-Nemotron-1.1-7Boriginalbf1616.098nvidia2025
agentflow-planner-7b-GGUFggufgguf16.036Mungert2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.