HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen3-1.7B-Base

Qwen/Qwen3-1.7B-Base

family qwen · 15 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen3-1.7B-Base
best scores
MMBench: 74.3ChartQA: 81.9DocVQA: 90.6MMLU: 56.1MMLU-Pro: 30.3HellaSwag: 59.4

Best scores per benchmark

benchmarkcategorybest score
DocVQAmultimodal90.6
ChartQAmultimodal81.9
MMBenchmultimodal74.3
HellaSwagknowledge59.4
MMLUknowledge56.1
MMLU-Proknowledge30.3

Variants (15)

variantkindquantbitsvRAMdlorgyear
Qwen3-1.7Boriginalbf1616.07,055,210Qwen2025
Qwen3-1.7B-Baseoriginalbf1616.0829,570Qwen2025
Qwen3-1.7B-Base-unsloth-bnb-4bitbnb-4bit4bit6,874unsloth2025
Qwen3-1.7B-Baseoriginalbf1616.02,062unsloth2025
Qwen3-1.7B-MLX-4bitmlxmlx719Qwen2025
MioTTS-1.7Boriginalbf1616.0628Aratako2026
Qwen3-1.7B-MLX-8bitmlxmlx573Qwen2025
jina-vlmoriginalbf1616.0529jinaai2026
MOSS-TTSD-v0.5originalbf1616.0455OpenMOSS-Team2025
Qwen3-1.7B-Base-bnb-4bitbnb-4bit4bit221unsloth2025
Qwen3-1.7B-MLX-bf16mlxmlx126Qwen2025
SAE-Res-Qwen3-1.7B-Base-W32K-L0_100originalbf1616.0103Qwen2026
Qwen3-1.7B-GGUFggufgguf16.087Mungert2025
Qwen3-1.7B-MLX-6bitmlxmlx71Qwen2025
SAE-Res-Qwen3-1.7B-Base-W32K-L0_50originalbf1616.064Qwen2026
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.