# MiMo-V2-Omni vs Qwen2.5 7B Instruct

> MiMo-V2-Omni is the stronger model overall, scoring 42.9 to 28.6 on the Noometry Index.

- Canonical page: https://noometry.com/compare/mimo-v2-omni-vs-qwen2-5-7b-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 0

## Summary

- The widest gap is in math, where MiMo-V2-Omni leads 36.4 to 11.8.
- MiMo-V2-Omni is cheaper at $0.14 / $0.28 per million input/output tokens, against $0.17 / $0.70 for Qwen2.5 7B Instruct.
- MiMo-V2-Omni accepts more context: 262K tokens versus 131K.
- Qwen2.5 7B Instruct has downloadable open weights; the other is API-only.

## Snapshot

| | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| Provider | Xiaomi | Alibaba (Qwen) |
| Noometry Index | 42.9 | 28.6 |
| Rank | 91 | 314 |
| Context | 262K | 131K |
| Input $/M | $0.14 | $0.17 |
| Output $/M | $0.28 | $0.70 |
| Weights | Proprietary | Open |

## Coding

- MiMo-V2-Omni: 42.7 (#87)
- Qwen2.5 7B Instruct: 35.9 (#204)

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| BigCodeBench Instruct | — | 37.6% |
| LMArena Coding | 1467 | — |
| BigCodeBench Complete | — | 46.1% |

## Agentic & Tool Use

- MiMo-V2-Omni: —
- Qwen2.5 7B Instruct: 23.5 (#117)

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| BALROG | — | 7.8% |

## Reasoning

- MiMo-V2-Omni: 28.6 (#86)
- Qwen2.5 7B Instruct: 14.3 (#317)

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| Chess Puzzles | — | 0% |
| LMArena Hard Prompts | 1445 | — |
| DTBench | — | 47.7% |
| LMCA | — | 6.4% |
| Epoch Capabilities Index | — | 118.44 |

## Math

- MiMo-V2-Omni: 36.4 (#112)
- Qwen2.5 7B Instruct: 11.8 (#300)

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 2.5% |
| Omni-MATH | — | 29.4% |
| LMArena Math | 1432 | — |

## Knowledge

- MiMo-V2-Omni: 39.9 (#114)
- Qwen2.5 7B Instruct: 17.0 (#281)

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| GPQA Diamond | — | 35.5% |
| MMLU-Pro | — | 53.9% |
| GPQA (HELM) | — | 34.1% |
| LMArena Expert | 1449 | — |
| MMLU | — | 72.9% |

## Multimodal

- MiMo-V2-Omni: 37.9 (#62)
- Qwen2.5 7B Instruct: —

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Vision | 1228 | — |

## Multilingual

- MiMo-V2-Omni: 52.0 (#97)
- Qwen2.5 7B Instruct: —

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Non-English | 1404 | — |
| LMArena Chinese | 1469 | — |
| LMArena French | 1448 | — |
| LMArena German | 1398 | — |
| LMArena Japanese | 1318 | — |
| LMArena Korean | 1355 | — |
| LMArena Russian | 1412 | — |
| LMArena Spanish | 1432 | — |

## Instruction Following

- MiMo-V2-Omni: 75.4 (#62)
- Qwen2.5 7B Instruct: 63.4 (#226)

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| IFEval | — | 74.1% |
| LMArena Instruction Following | 1428 | — |

## Long Context

- MiMo-V2-Omni: 44.1 (#72)
- Qwen2.5 7B Instruct: —

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Longer Query | 1442 | — |

## Writing & Preference

- MiMo-V2-Omni: 61.6 (#83)
- Qwen2.5 7B Instruct: 48.8 (#194)

| Benchmark | MiMo-V2-Omni | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Text | 1424 | — |
| LMArena Creative Writing | 1393 | — |
| WildBench | — | 73.1% |
| LMArena Multi-Turn | 1446 | — |

## FAQ

### Is MiMo-V2-Omni better than Qwen2.5 7B Instruct?

MiMo-V2-Omni is the stronger model overall, scoring 42.9 to 28.6 on the Noometry Index.

### Which is cheaper, MiMo-V2-Omni or Qwen2.5 7B Instruct?

MiMo-V2-Omni is cheaper. It lists at $0.14 per million input tokens and $0.28 per million output tokens; Qwen2.5 7B Instruct lists at $0.17 and $0.70.

### Is MiMo-V2-Omni or Qwen2.5 7B Instruct better for coding?

MiMo-V2-Omni scores higher on coding benchmarks: 42.7 versus 35.9 in the Noometry coding category.

### Which has the bigger context window?

MiMo-V2-Omni does, with 262K tokens against 131K.

### How many benchmarks do MiMo-V2-Omni and Qwen2.5 7B Instruct share?

0 benchmarks have published results for both models. MiMo-V2-Omni has 18 scored results on Noometry and Qwen2.5 7B Instruct has 15.