Agentic Information Retrieval on BrowseComp-Plus 100k corpus (100-query sample)
82AccuracyRISE
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| RISEModel=gpt-5.42026.06 | 82 | 8 | 32.2 | 11 | 16.2 | 7.1 | 88.5 | 92.4 | 1.25 | |
| RISEModel=mimo-v2.5-pro2026.06 | 78 | 2 | 23.5 | 16.2 | 4.8 | 7.9 | 79.4 | 80.2 | 0.38 | |
| RISEModel=gpt-5.4-mini (medium effort)2026.06 | 78 | 1 | 24.3 | 13.1 | 9.2 | 6.4 | 78.5 | 79.6 | 0.28 | |
| DCIModel=gpt-5.4-mini (medium effort)2026.06 | 78 | 5 | 48.8 | — | 90.3 | 8.8 | — | 82.4 | 1.1 | |
| RISE-BM25Model=gpt-5.4-mini (medium effort)2026.06 | 77 | 1 | 23 | 14.8 | 9.6 | 5.2 | 77.7 | 77.9 | 0.34 | |
| RISE-BM25Model=mimo-v2.5-pro2026.06 | 75 | 1 | 19.8 | 15.7 | 3.6 | 6.5 | 75.7 | 75.7 | 0.37 | |
| retrieval-agentModel=mimo-v2.5-pro2026.06 | 73 | 1 | 34.5 | 30.8 | — | 2.5 | 76.5 | 76.5 | 0.61 | |
| DCIModel=gpt-5.4-nano (high effort)2026.06 | 71 | 7 | 45.7 | — | 119.4 | 7.1 | — | 77.4 | 0.2 | |
| retrieval-agentModel=gpt-5.4-mini (medium effort)2026.06 | 68 | 12 | 29.2 | 37 | — | 1.9 | 77.5 | 77.5 | 0.46 | |
| RISEModel=gpt-5.4-nano (high effort)2026.06 | 68 | 0 | 24.2 | 10.4 | 8.4 | 4.6 | 73.4 | 74.8 | 0.05 | |
| retrieval-agentModel=gpt-5.4-nano (high effort)2026.06 | 65 | 0 | 17 | 14.3 | — | 2 | 62 | 62 | 0.06 | |
| RISE-BM25Model=gpt-5.4-nano (high effort)2026.06 | 64 | 0 | 22.9 | 10.2 | 8.1 | 3.7 | 72.2 | 75.4 | 0.06 | |
| DCIModel=mimo-v2.5-pro2026.06 | 60 | 18 | 26 | — | 51.8 | 5.5 | — | 62 | 0.52 |