Qwen
Benchmarks
TaskDatasetSOTA resultTrendResults
Qwen3-1.7B Inference Performance (test)
100High-Precision Ratio
7
Qwen3-4B Inference Performance (test)
100High-Precision Ratio
7
Qwen3-8B Inference Performance (test)
100High-Precision Ratio
7
Qwen 7B 2.5
6.38Avg VRAM (MB)
7
Qwen 7B 2.5
22.2End Pass@1
6
Qwen 3.5 (val)
8.89Perplexity (PPL)
6
Qwen3-8B
2Score
6
Qwen2.5-7B Adaptive PAIR attack
4ASR
6
Qwen3-0.6B (inference)
0.255Storage (GB)
6
Qwen 0.5B Instruct 2.5
185.5Throughput (Tok/s)
6
Qwen2.5-14B-Instruct High-Variance (Top 20%)
5.67Average Reward (μ)
6
Qwen2.5-14B-Instruct Overall
6.31Reward (Avg μ)
6
Qwen VL 2.5
87KMR (a)
6
Qwen (test)
22.54Throughput (tokens/s)
6
Qwen 7B Uniform distribution 32 model variants 2.5
0.86Throughput (req/s)
6
Qwen 7B (Zipf distribution, alpha=1.5) 2.5 (32 model variants)
0.87Throughput (req/s)
6
Qwen3-8B serving environment
0.12TTFT (s)
6
Qwen2.5-14B Wealth-Seeking
67.5Wealth-Seeking Score
6
Qwen2.5-14B-instruct (test)
8,192Response Length
6
Qwen
100Detectability (Origin)
5
Qwen3-8B
54.2User Throughput (tokens/s)
5
Qwen3-4B-Thinking-2507
66.4Throughput (tokens/s/user)
5
Qwen3-8B
196.8Throughput (tokens/sec)
5
Qwen3-4B
231.3Throughput (tokens/sec)
5
Qwen3 1.7B
316.9Decoding Throughput (tokens/sec)
5