# MiMo v2.5 · Search Capability Leaderboard

* Search Intelligence Score: **45.0** (#20 in Search Intelligence)
* Without search: 12.7 · Lift from search +32.3
* Cost per 1K tasks: $26.2 (below the median score of 51.2, so not ranked on cost)
* Time per task: 203s
* Lab: Xiaomi · Model id: xiaomi/mimo-v2.5

## Accuracy by benchmark

| Suite | With search | Without search |
| ----- | ----------- | -------------- |
| DSQA  | 40.0        | 10.0           |
| HLE   | 49.0        | 21.0           |
| WISER | 46.0        | 7.0            |

## Compared with

| Model                                                            | Score | Lift  | Cost per 1K tasks | Time per task |
| ---------------------------------------------------------------- | ----- | ----- | ----------------- | ------------- |
| MiMo v2.5                                                        | 45.0  | +32.3 | $26.2             | 203s          |
| [Gemini 3 Flash](/leaderboard/gemini-3-flash-preview)            | 45.3  | +21.0 | $199              | 116s          |
| [DeepSeek V4 Flash (0731)](/leaderboard/deepseek-v4-flash-0731)  | 43.7  | +21.0 | $9.45             | 88.0s         |
| [Nemotron 3 Ultra 550B](/leaderboard/nemotron-3-ultra-550b-a55b) | 42.7  | +26.3 | $155              | 109s          |
| [MiniMax M3](/leaderboard/minimax-m3)                            | 49.0  | +29.0 | $39.9             | 156s          |

Latest update September 14, 2026\. Methodology: /leaderboard#methodology