# Qwen3 Family Inference Performance (PyTorch Workflow) Source: [Qwen3 Family Inference Performance (PyTorch Workflow)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:52:55.279Z Evaluation of inference performance metrics (load time, latency, throughput, and memory usage) for models within the Qwen3 family when executed using a standard Transformers (PyTorch) workflow. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | LLM Inference | | Dataset (including any stated split) | Transformers (PyTorch) workflow Qwen3 family (inference) | | Metric | Model Load Time (s) | | Unit | TIME | | Direction | Lower is better | | Results loaded | 18 of 18 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Model Load Time (s)](): Lower is better; unit: TIME. - [First Token Latency (s)](): Lower is better; unit: TIME. - [Throughput (tok/s)](): Higher is better; unit: SCALAR. - [CPU Memory (MiB)](): Lower is better; unit: SIZE. - [Accelerator Memory (GiB)](): Lower is better; unit: SIZE. ## Recorded results | Method | Model Load Time (s) | Result date | Report paper | | --- | --- | --- | --- | | CryptoTensors | 1.16 | Not recorded | 693266f0d6b0dd334293419d | | Safetensors | 1.169 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 1.699 | Not recorded | 693266f0d6b0dd334293419d | | Safetensors | 1.76 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 2.613 | Not recorded | 693266f0d6b0dd334293419d | | Safetensors | 2.649 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 2.713 | Not recorded | 693266f0d6b0dd334293419d | | Safetensors | 4.563 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 4.674 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 4.747 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 9.388 | Not recorded | 693266f0d6b0dd334293419d | | Safetensors | 9.558 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 10.552 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 20.365 | Not recorded | 693266f0d6b0dd334293419d | | Safetensors | 21.227 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 25.569 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 50.171 | Not recorded | 693266f0d6b0dd334293419d | | CryptoTensors | 107.714 | Not recorded | 693266f0d6b0dd334293419d | ### CryptoTensors Model Load Time (s): 1.16. Result ID: 6997086a5698665ec455b7aa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-0.6B","Configuration":"CryptoTensors-Unencrypted"}. This record does not assert independent reproduction. ### Safetensors Model Load Time (s): 1.169. Result ID: 6997086a5698665ec455b7a9. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-0.6B","Configuration":"Safetensors"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 1.699. Result ID: 6997086a5698665ec455b7ad. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-1.7B","Configuration":"CryptoTensors-Unencrypted"}. This record does not assert independent reproduction. ### Safetensors Model Load Time (s): 1.76. Result ID: 6997086a5698665ec455b7ac. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-1.7B","Configuration":"Safetensors"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 2.613. Result ID: 6997086a5698665ec455b7b0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Configuration":"CryptoTensors-Unencrypted"}. This record does not assert independent reproduction. ### Safetensors Model Load Time (s): 2.649. Result ID: 6997086a5698665ec455b7af. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Configuration":"Safetensors"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 2.713. Result ID: 6997086a5698665ec455b7ab. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-0.6B","Configuration":"CryptoTensors-Encrypted"}. This record does not assert independent reproduction. ### Safetensors Model Load Time (s): 4.563. Result ID: 6997086a5698665ec455b7b2. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-8B","Configuration":"Safetensors"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 4.674. Result ID: 6997086a5698665ec455b7b3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-8B","Configuration":"CryptoTensors-Unencrypted"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 4.747. Result ID: 6997086a5698665ec455b7ae. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-1.7B","Configuration":"CryptoTensors-Encrypted"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 9.388. Result ID: 6997086a5698665ec455b7b6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-14B","Configuration":"CryptoTensors-Unencrypted"}. This record does not assert independent reproduction. ### Safetensors Model Load Time (s): 9.558. Result ID: 6997086a5698665ec455b7b5. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-14B","Configuration":"Safetensors"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 10.552. Result ID: 6997086a5698665ec455b7b1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Configuration":"CryptoTensors-Encrypted"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 20.365. Result ID: 6997086a5698665ec455b7b9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-32B","Configuration":"CryptoTensors-Unencrypted"}. This record does not assert independent reproduction. ### Safetensors Model Load Time (s): 21.227. Result ID: 6997086a5698665ec455b7b8. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-32B","Configuration":"Safetensors"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 25.569. Result ID: 6997086a5698665ec455b7b4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-8B","Configuration":"CryptoTensors-Encrypted"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 50.171. Result ID: 6997086a5698665ec455b7b7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-14B","Configuration":"CryptoTensors-Encrypted"}. This record does not assert independent reproduction. ### CryptoTensors Model Load Time (s): 107.714. Result ID: 6997086a5698665ec455b7ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-32B","Configuration":"CryptoTensors-Encrypted"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.