vLLM Inference Performance on Qwen3-4B
1.101Model Load Time (s)Safetensors
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Safetensors2025.12 | 1.101 | 123.277 | 126.13 | 11,075.59 | 7.58 | |
| CryptoTensorsEncryption=Unencrypted2025.12 | 1.104 | 123.406 | 125.23 | 11,088.67 | 7.58 | |
| CryptoTensorsEncryption=Encrypted2025.12 | 11.918 | 123.467 | 126.84 | 11,608.82 | 7.58 |