# Long-Context Generation Efficiency on Qwen2.5-72B

Source: [Long-Context Generation Efficiency on Qwen2.5-72B](<https://www.sota2.com/research/sota/long-context-generation-efficiency-on-qwen2-5-72b>)

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

Data updated: 2026-06-25T07:36:32.047Z

Evaluation of the long-context generation efficiency for the Qwen2.5-72B-Instruct model, specifically measuring speedup performance across context lengths of 8K, 16K, and 32K tokens.

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 | Long-Context Generation Efficiency |
| Dataset (including any stated split) | Qwen2.5-72B-Instruct Long-Context Efficiency Workload |
| Metric | Speedup (8K) |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| 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

- [Speedup (8K)](<https://www.sota2.com/research/sota/long-context-generation-efficiency-on-qwen2-5-72b.md?metric=Speedup%20(8K)>): Higher is better; unit: SCALAR.

- [Speedup (16K)](<https://www.sota2.com/research/sota/long-context-generation-efficiency-on-qwen2-5-72b.md?metric=Speedup%20(16K)>): Higher is better; unit: SCALAR.

- [Speedup (32K)](<https://www.sota2.com/research/sota/long-context-generation-efficiency-on-qwen2-5-72b.md?metric=Speedup%20(32K)>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Speedup (8K) | Result date | Report paper |
| --- | --- | --- | --- |
| Dustin | 4.76 | Not recorded | 6a3cb5fb206542251d7ca492 |
| SpecAttn | 3.91 | Not recorded | 6a3cb5fb206542251d7ca492 |

### Dustin

Speedup (8K): 4.76. Result ID: 6a3cda71fbc373592c517fcb.

[Reporting paper](<https://www.sota2.com/research/paper/6a3cb5fb206542251d7ca492.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3cb5fb206542251d7ca492.md>)

Recorded evaluation configuration: {"Batch Size":"16"}. This record does not assert independent reproduction.

### SpecAttn

Speedup (8K): 3.91. Result ID: 6a3cda71fbc373592c517fca.

[Reporting paper](<https://www.sota2.com/research/paper/6a3cb5fb206542251d7ca492.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690836e63ef7ccd74e5e5301.md>)

Recorded evaluation configuration: {"Batch Size":"16"}. This record does not assert independent reproduction.

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