# vLLM Inference Performance on Qwen3-14B

Source: [vLLM Inference Performance on Qwen3-14B](<https://www.sota2.com/research/sota/vllm-inference-performance-on-qwen3-14b>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:52:55.247Z

Evaluation of the Qwen3-14B model's inference performance and resource utilization when deployed using the vLLM framework, covering load time, latency, throughput, and memory consumption.

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 | vLLM Model Deployment and Inference |
| Dataset (including any stated split) | Qwen3-14B vLLM (inference) |
| Metric | Model Load Time (s) |
| Unit | TIME |
| Direction | Lower is better |
| Results loaded | 3 of 3 |
| 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)](<https://www.sota2.com/research/sota/vllm-inference-performance-on-qwen3-14b.md?metric=Model%20Load%20Time%20(s)>): Lower is better; unit: TIME.

- [First Token Latency (s)](<https://www.sota2.com/research/sota/vllm-inference-performance-on-qwen3-14b.md?metric=First%20Token%20Latency%20(s)>): Lower is better; unit: TIME.

- [Throughput (tok/s)](<https://www.sota2.com/research/sota/vllm-inference-performance-on-qwen3-14b.md?metric=Throughput%20(tok%2Fs)>): Higher is better; unit: SCALAR.

- [CPU Memory (MiB)](<https://www.sota2.com/research/sota/vllm-inference-performance-on-qwen3-14b.md?metric=CPU%20Memory%20(MiB)>): Lower is better; unit: SIZE.

- [Accelerator Memory (GiB)](<https://www.sota2.com/research/sota/vllm-inference-performance-on-qwen3-14b.md?metric=Accelerator%20Memory%20(GiB)>): Lower is better; unit: SIZE.

## Recorded results

| Method | Model Load Time (s) | Result date | Report paper |
| --- | --- | --- | --- |
| Safetensors | 3.082 | Not recorded | 693266f0d6b0dd334293419d |
| CryptoTensors | 3.165 | Not recorded | 693266f0d6b0dd334293419d |
| CryptoTensors | 61.879 | Not recorded | 693266f0d6b0dd334293419d |

### Safetensors

Model Load Time (s): 3.082. Result ID: 6997086a5698665ec455b7a3.

[Reporting paper](<https://www.sota2.com/research/paper/693266f0d6b0dd334293419d.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### CryptoTensors

Model Load Time (s): 3.165. Result ID: 6997086a5698665ec455b7a4.

[Reporting paper](<https://www.sota2.com/research/paper/693266f0d6b0dd334293419d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693266f0d6b0dd334293419d.md>)

Recorded evaluation configuration: {"Encryption":"Unencrypted"}. This record does not assert independent reproduction.

### CryptoTensors

Model Load Time (s): 61.879. Result ID: 6997086a5698665ec455b7a5.

[Reporting paper](<https://www.sota2.com/research/paper/693266f0d6b0dd334293419d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693266f0d6b0dd334293419d.md>)

Recorded evaluation configuration: {"Encryption":"Encrypted"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.