# Long-context understanding on LongBench 32k 2024

Source: [Long-context understanding on LongBench 32k 2024](<https://www.sota2.com/research/sota/long-context-understanding-on-longbench-32k-2024>)

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

Data updated: 2026-05-11T09:24:08.276Z

Evaluation of model performance on the LongBench suite specifically configured for the 32k context length version released in 2024, focusing on key QA and reading comprehension subtasks.

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 understanding |
| Dataset (including any stated split) | LongBench 32k 2024 |
| Metric | HotpotQA Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [HotpotQA Score](<https://www.sota2.com/research/sota/long-context-understanding-on-longbench-32k-2024.md?metric=HotpotQA%20Score>): Higher is better; unit: PERCENT.

- [NarrativeQA Score](<https://www.sota2.com/research/sota/long-context-understanding-on-longbench-32k-2024.md?metric=NarrativeQA%20Score>): Higher is better; unit: PERCENT.

- [TriviaQA Score](<https://www.sota2.com/research/sota/long-context-understanding-on-longbench-32k-2024.md?metric=TriviaQA%20Score>): Higher is better; unit: PERCENT.

- [TREC Score](<https://www.sota2.com/research/sota/long-context-understanding-on-longbench-32k-2024.md?metric=TREC%20Score>): Higher is better; unit: PERCENT.

- [MultiNews Score](<https://www.sota2.com/research/sota/long-context-understanding-on-longbench-32k-2024.md?metric=MultiNews%20Score>): Higher is better; unit: PERCENT.

- [Qasper Score](<https://www.sota2.com/research/sota/long-context-understanding-on-longbench-32k-2024.md?metric=Qasper%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | HotpotQA Score | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen2.5 | 57.62 | Not recorded | 68fa7c421e209131cea8d35b |
| SpikingBrain-76B | 50 | Not recorded | 68fa7c421e209131cea8d35b |
| Mixtral | 36.42 | Not recorded | 68fa7c421e209131cea8d35b |
| Jamba | 11.38 | Not recorded | 68fa7c421e209131cea8d35b |

### Qwen2.5

HotpotQA Score: 57.62. Result ID: 6a01a01d47f34ce924e3c94f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c421e209131cea8d35b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa767e1e209131cea6d6a7.md>)

Recorded evaluation configuration: {"Params":"7B","Complexity Type":"Quadratic","Context Truncation":"32k"}. This record does not assert independent reproduction.

### SpikingBrain-76B

HotpotQA Score: 50. Result ID: 6a01a01d47f34ce924e3c94d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c421e209131cea8d35b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c421e209131cea8d35b.md>)

Recorded evaluation configuration: {"Params":"12B/76B","Complexity Type":"Hybrid","Context Truncation":"32k"}. This record does not assert independent reproduction.

### Mixtral

HotpotQA Score: 36.42. Result ID: 6a01a01d47f34ce924e3c950.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c421e209131cea8d35b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71d01e209131cea5317b.md>)

[Implementation](<https://github.com/mistralai/mistral-src>)

Recorded evaluation configuration: {"Params":"13B/47B","Complexity Type":"Quadratic","Context Truncation":"32k"}. This record does not assert independent reproduction.

### Jamba

HotpotQA Score: 11.38. Result ID: 6a01a01d47f34ce924e3c94e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7c421e209131cea8d35b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74061e209131cea5f4f6.md>)

Recorded evaluation configuration: {"Params":"12B/52B","Complexity Type":"Hybrid","Context Truncation":"32k"}. This record does not assert independent reproduction.

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