# End-to-end Retrieval on MSMARCO (R@10, R@100, R@1k, Latency)

Source: [End-to-end Retrieval on MSMARCO (R@10, R@100, R@1k, Latency)](<https://www.sota2.com/research/sota/end-to-end-retrieval-on-msmarco-r-at-10-r-at-100-r-at-1k-latency>)

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

Data updated: 2026-03-24T12:54:44.138Z

Evaluation of end-to-end retrieval performance on the MSMARCO dataset, reporting Recall metrics at top 10, 100, and 1k, alongside latency measurements in milliseconds.

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 | End-to-end Retrieval |
| Dataset (including any stated split) | MSMARCO |
| Metric | Latency (ms) |
| 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

- [Latency (ms)](<https://www.sota2.com/research/sota/end-to-end-retrieval-on-msmarco-r-at-10-r-at-100-r-at-1k-latency.md?metric=Latency%20(ms)>): Lower is better; unit: TIME.

- [Recall@10](<https://www.sota2.com/research/sota/end-to-end-retrieval-on-msmarco-r-at-10-r-at-100-r-at-1k-latency.md?metric=Recall%4010>): Higher is better; unit: PERCENT.

- [Recall@100](<https://www.sota2.com/research/sota/end-to-end-retrieval-on-msmarco-r-at-10-r-at-100-r-at-1k-latency.md?metric=Recall%40100>): Higher is better; unit: PERCENT.

- [Recall@1k](<https://www.sota2.com/research/sota/end-to-end-retrieval-on-msmarco-r-at-10-r-at-100-r-at-1k-latency.md?metric=Recall%401k>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Latency (ms) | Result date | Report paper |
| --- | --- | --- | --- |
| GEM | 88 | Not recorded | 69c21a890636e5bf4afc482c |
| GEM | 140 | Not recorded | 69c21a890636e5bf4afc482c |
| GEM | 212 | Not recorded | 69c21a890636e5bf4afc482c |
| PLAID | 222 | Not recorded | 69c21a890636e5bf4afc482c |
| IGP | 279 | Not recorded | 69c21a890636e5bf4afc482c |
| MUVERA | 286 | Not recorded | 69c21a890636e5bf4afc482c |
| MVG | 293 | Not recorded | 69c21a890636e5bf4afc482c |
| DESSERT | 298 | Not recorded | 69c21a890636e5bf4afc482c |
| MUVERA | 310 | Not recorded | 69c21a890636e5bf4afc482c |
| DESSERT | 336 | Not recorded | 69c21a890636e5bf4afc482c |
| IGP | 340 | Not recorded | 69c21a890636e5bf4afc482c |
| DESSERT | 344 | Not recorded | 69c21a890636e5bf4afc482c |
| PLAID | 352 | Not recorded | 69c21a890636e5bf4afc482c |
| PLAID | 352 | Not recorded | 69c21a890636e5bf4afc482c |
| MVG | 360 | Not recorded | 69c21a890636e5bf4afc482c |
| IGP | 399 | Not recorded | 69c21a890636e5bf4afc482c |
| MVG | 436 | Not recorded | 69c21a890636e5bf4afc482c |
| MUVERA | 444 | Not recorded | 69c21a890636e5bf4afc482c |

### GEM

Latency (ms): 88. Result ID: 69c288dc47f34ce924aa9b33.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"10","ef\_search":"800"}. This record does not assert independent reproduction.

### GEM

Latency (ms): 140. Result ID: 69c288dc47f34ce924aa9b34.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"100","ef\_search":"2000"}. This record does not assert independent reproduction.

### GEM

Latency (ms): 212. Result ID: 69c288dc47f34ce924aa9b35.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"1000","ef\_search":"8000"}. This record does not assert independent reproduction.

### PLAID

Latency (ms): 222. Result ID: 69c288dc47f34ce924aa9b24.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c9e1e209131cea34fac.md>)

Recorded evaluation configuration: {"k (number of returned results)":"10"}. This record does not assert independent reproduction.

### IGP

Latency (ms): 279. Result ID: 69c288dc47f34ce924aa9b2d.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"10"}. This record does not assert independent reproduction.

### MUVERA

Latency (ms): 286. Result ID: 69c288dc47f34ce924aa9b2a.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa736d1e209131cea5c6f6.md>)

Recorded evaluation configuration: {"k (number of returned results)":"10"}. This record does not assert independent reproduction.

### MVG

Latency (ms): 293. Result ID: 69c288dc47f34ce924aa9b30.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"10"}. This record does not assert independent reproduction.

### DESSERT

Latency (ms): 298. Result ID: 69c288dc47f34ce924aa9b27.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71071e209131cea4eb21.md>)

[Implementation](<https://github.com/ThirdAIResearch/Dessert>)

Recorded evaluation configuration: {"k (number of returned results)":"10"}. This record does not assert independent reproduction.

### MUVERA

Latency (ms): 310. Result ID: 69c288dc47f34ce924aa9b2b.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa736d1e209131cea5c6f6.md>)

Recorded evaluation configuration: {"k (number of returned results)":"100"}. This record does not assert independent reproduction.

### DESSERT

Latency (ms): 336. Result ID: 69c288dc47f34ce924aa9b29.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71071e209131cea4eb21.md>)

[Implementation](<https://github.com/ThirdAIResearch/Dessert>)

Recorded evaluation configuration: {"k (number of returned results)":"1000"}. This record does not assert independent reproduction.

### IGP

Latency (ms): 340. Result ID: 69c288dc47f34ce924aa9b2e.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"100"}. This record does not assert independent reproduction.

### DESSERT

Latency (ms): 344. Result ID: 69c288dc47f34ce924aa9b28.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71071e209131cea4eb21.md>)

[Implementation](<https://github.com/ThirdAIResearch/Dessert>)

Recorded evaluation configuration: {"k (number of returned results)":"100"}. This record does not assert independent reproduction.

### PLAID

Latency (ms): 352. Result ID: 69c288dc47f34ce924aa9b25.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c9e1e209131cea34fac.md>)

Recorded evaluation configuration: {"k (number of returned results)":"100"}. This record does not assert independent reproduction.

### PLAID

Latency (ms): 352. Result ID: 69c288dc47f34ce924aa9b26.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c9e1e209131cea34fac.md>)

Recorded evaluation configuration: {"k (number of returned results)":"1000"}. This record does not assert independent reproduction.

### MVG

Latency (ms): 360. Result ID: 69c288dc47f34ce924aa9b31.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"100"}. This record does not assert independent reproduction.

### IGP

Latency (ms): 399. Result ID: 69c288dc47f34ce924aa9b2f.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"1000"}. This record does not assert independent reproduction.

### MVG

Latency (ms): 436. Result ID: 69c288dc47f34ce924aa9b32.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

Recorded evaluation configuration: {"k (number of returned results)":"1000"}. This record does not assert independent reproduction.

### MUVERA

Latency (ms): 444. Result ID: 69c288dc47f34ce924aa9b2c.

[Reporting paper](<https://www.sota2.com/research/paper/69c21a890636e5bf4afc482c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa736d1e209131cea5c6f6.md>)

Recorded evaluation configuration: {"k (number of returned results)":"1000"}. This record does not assert independent reproduction.

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