HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen3-8B

Qwen/Qwen3-8B

family qwen · 44 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen3-8B
best scores
GPQA: 62.0MMLU: 87.5AIME 2024: 76.0WebArena: 26.6RULER: 69.2GPQA Diamond: 58.8MMLU-Pro: 74.2MATH-500: 95.1HumanEval: 87.8IFEval: 90.3Chatbot Arena: 88.4

Best scores per benchmark

benchmarkcategorybest score
MATH-500math95.1
IFEvalinstruction-following90.3
Chatbot Arenachat88.4
HumanEvalcoding87.8
MMLUknowledge87.5
AIME 2024math76.0
MMLU-Proknowledge74.2
RULERlong-context69.2
GPQAreasoning62.0
GPQA Diamondreasoning58.8
WebArenaagentic26.6

Variants (44)

variantkindquantbitsvRAMdlorgyear
Qwen3-8B-FP8fp8fp88.0835,337Qwen2025
Qwen3-8B-AWQawqawq824,569Qwen2025
Qwen3-8B-GGUFggufgguf16.0323,626unsloth2025
Molmo2-8Boriginalbf1616.0252,079allenai2026
Qwen3-8B-GGUFggufgguf4.0178,323Qwen2025
Qwen3-8B-NVFP4nvfp4nvfp44.0159,182nvidia2025
Qwen3-8B-unsloth-bnb-4bitbnb-4bit4bit149,869unsloth2025
Qwen3-8Boriginalbf1616.0123,753unsloth2025
Qwen3Guard-Gen-8Boriginalbf1616.050,770Qwen2025
Qwen3-8B-MLX-4bitmlxmlx28,941lmstudio-community2025
Qwen3-8B-MLX-8bitmlxmlx27,769lmstudio-community2025
Qwen3-8B-GGUFggufgguf3.024,231lmstudio-community2025
Qwen3-8B-bnb-4bitbnb-4bit4bit21,474unsloth2025
Qwen_Qwen3-8B-GGUFggufgguf2.021,193bartowski2025
MolmoWeb-4Boriginalbf1616.06,233allenai2026
Qwen3-8B-FP8fp8fp88.04,389nvidia2025
Qwen3-8B-abliterated-GGUFggufgguf16.03,611Mungert2025
MolmoPoint-8Boriginalbf1616.03,601allenai2026
Nemotron-Orchestrator-8Boriginalbf1616.02,271nvidia2025
Qwen3-8B-128K-GGUFggufgguf16.01,912unsloth2025
Qwen3-8B-GGUFggufgguf4.01,888ggml-org2025
tmax-sft-8boriginalbf1616.01,797allenai2026
Qwen3Guard-Stream-8Boriginalbf1616.01,753Qwen2025
Qwen3-8B-MLX-4bitmlxmlx1,691Qwen2025
Qwen3-8B-i1-GGUFggufgguf1.01,465mradermacher2025
MolmoWeb-8Boriginalbf1616.0931allenai2026
WebWorld-8Boriginalbf1616.0709Qwen2026
Qwen3-8B-MLX-8bitmlxmlx574Qwen2025
Qwen3-8B-abliteratedabliteratedabliterated526mlabonne2025
WebWorld-14Boriginalbf1616.0474Qwen2026
Nemotron-Orchestrator-8B-GGUFggufgguf16.0394Mungert2025
MiroThinker-v1.0-8B-GGUFggufgguf16.0352Mungert2025
Qwen3-Nemotron-8B-BRRMoriginalbf1616.0351nvidia2025
Josiefied-Qwen3-8B-abliterated-v1-GGUFggufgguf16.0351Mungert2025
WebWorld-32Boriginalbf1616.0348Qwen2026
Qwen3-8B-DMS-8xoriginalbf1616.0238nvidia2026
Qwen3-8B-MLX-bf16mlxmlx145Qwen2025
UIGEN-X-8B-GGUFggufgguf16.0122Mungert2025
Qwen3-8B-MLX-6bitmlxmlx103Qwen2025
intent-aware-lfqa-qwen3-8b-baselineoriginalbf1616.063allenai2026
intent-aware-lfqa-qwen3-8b-intent-implicitoriginalbf1616.061allenai2026
intent-aware-lfqa-qwen3-8b-multivieworiginalbf1616.060allenai2026
intent-aware-lfqa-qwen3-8b-intent-explicitoriginalbf1616.058allenai2026
Qwen3Guard-Gen-8B-GGUFggufgguf16.027Mungert2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.