vLLM Inference Performance on Qwen3-8B
1.732Model Load Time (s)Safetensors
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Safetensors2025.12 | 1.732 | 122.039 | 115.87 | 11,095.98 | 15.3 | |
| CryptoTensorsEncryption=Unencrypted2025.12 | 1.806 | 122.563 | 115.61 | 11,108.98 | 15.3 | |
| CryptoTensorsEncryption=Encrypted2025.12 | 27.056 | 123.134 | 112.85 | 11,233.99 | 15.3 |