# MiniMax-M3 vs Qwen2.5 7B Instruct

> MiniMax-M3 is the stronger model overall, scoring 43.5 to 28.6 on the Noometry Index. Qwen2.5 7B Instruct costs 1.7× less per token, which makes it the better buy when MiniMax-M3's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/minimax-m3-vs-qwen2-5-7b-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 6

## Summary

- They share 6 benchmarks with published results for both. MiniMax-M3 scores higher in 6 categories and Qwen2.5 7B Instruct in 1 category; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where MiniMax-M3 leads 58.4 to 17.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 71.1% for MiniMax-M3 and 2.5% for Qwen2.5 7B Instruct.
- Qwen2.5 7B Instruct is cheaper at $0.17 / $0.70 per million input/output tokens, against $0.30 / $1.20 for MiniMax-M3.
- MiniMax-M3 accepts more context: 1M tokens versus 131K.

## Snapshot

| | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| Provider | MiniMax | Alibaba (Qwen) |
| Noometry Index | 43.5 | 28.6 |
| Rank | 81 | 314 |
| Context | 1M | 131K |
| Input $/M | $0.30 | $0.17 |
| Output $/M | $1.20 | $0.70 |
| Weights | Open | Open |

## Coding

- MiniMax-M3: 41.5 (#111)
- Qwen2.5 7B Instruct: 35.9 (#204)

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| FrontierCode | 14.7% | — |
| LMArena WebDev | 1482 | — |
| SciCode | 47.1% | — |
| BigCodeBench Instruct | — | 37.6% |
| LMArena Coding | 1470 | — |
| BigCodeBench Complete | — | 46.1% |
| ALE-Bench | 640.02 | — |

## Agentic & Tool Use

- MiniMax-M3: 22.5 (#121)
- Qwen2.5 7B Instruct: 23.5 (#117)

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| APEX-Agents | 37.7% | — |
| OSWorld 2.0 | 4.6% | — |
| BALROG | — | 7.8% |
| GBAEval | 0.9% | — |
| Vending-Bench 2 | 2,158 | — |

## Reasoning

- MiniMax-M3: 29.3 (#83)
- Qwen2.5 7B Instruct: 14.3 (#317)

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| Chess Puzzles | 14% | 0% |
| DTBench | 78.9% | 47.7% |
| LMCA | 33.7% | 6.4% |
| Epoch Capabilities Index | 147 | 118.44 |
| SimpleBench | 45.8% | — |
| NYT Connections (extended) | 65.1% | — |
| CritPt | 3.7% | — |
| LMArena Hard Prompts | 1447 | — |
| Mystery Game Puzzles | 8% | — |
| Surface Evolver Bench | 55% | — |
| ForecastBench | 61.4 | — |

## Math

- MiniMax-M3: 39.1 (#83)
- Qwen2.5 7B Instruct: 11.8 (#300)

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 2.5% |
| ProofBench | 18% | — |
| Omni-MATH | — | 29.4% |
| LMArena Math | 1432 | — |

## Knowledge

- MiniMax-M3: 58.4 (#32)
- Qwen2.5 7B Instruct: 17.0 (#281)

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| GPQA Diamond | 90.9% | 35.5% |
| MMLU-Pro | — | 53.9% |
| GPQA (HELM) | — | 34.1% |
| LMArena Expert | 1462 | — |
| MMLU | — | 72.9% |

## Multimodal

- MiniMax-M3: 39.5 (#45)
- Qwen2.5 7B Instruct: —

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Vision | 1254 | — |
| LMArena Document | 1435 | — |

## Multilingual

- MiniMax-M3: 53.2 (#70)
- Qwen2.5 7B Instruct: —

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Non-English | 1420 | — |
| LMArena Chinese | 1465 | — |
| LMArena French | 1445 | — |
| LMArena German | 1424 | — |
| LMArena Japanese | 1381 | — |
| LMArena Korean | 1372 | — |
| LMArena Russian | 1430 | — |
| LMArena Spanish | 1431 | — |

## Instruction Following

- MiniMax-M3: 75.7 (#56)
- Qwen2.5 7B Instruct: 63.4 (#226)

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| IFEval | — | 74.1% |
| LMArena Instruction Following | 1433 | — |

## Long Context

- MiniMax-M3: 44.2 (#68)
- Qwen2.5 7B Instruct: —

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Longer Query | 1445 | — |

## Writing & Preference

- MiniMax-M3: 62.3 (#79)
- Qwen2.5 7B Instruct: 48.8 (#194)

| Benchmark | MiniMax-M3 | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Text | 1433 | — |
| LMArena Creative Writing | 1405 | — |
| WildBench | — | 73.1% |
| EQ-Bench 4 | 1150 | — |
| LMArena Multi-Turn | 1441 | — |

## FAQ

### Is MiniMax-M3 better than Qwen2.5 7B Instruct?

MiniMax-M3 is the stronger model overall, scoring 43.5 to 28.6 on the Noometry Index. Qwen2.5 7B Instruct costs 1.7× less per token, which makes it the better buy when MiniMax-M3's lead doesn't matter for your workload.

### Which is cheaper, MiniMax-M3 or Qwen2.5 7B Instruct?

Qwen2.5 7B Instruct is cheaper. It lists at $0.17 per million input tokens and $0.70 per million output tokens; MiniMax-M3 lists at $0.30 and $1.20.

### Is MiniMax-M3 or Qwen2.5 7B Instruct better for coding?

MiniMax-M3 scores higher on coding benchmarks: 41.5 versus 35.9 in the Noometry coding category.

### Which has the bigger context window?

MiniMax-M3 does, with 1M tokens against 131K.

### How many benchmarks do MiniMax-M3 and Qwen2.5 7B Instruct share?

6 benchmarks have published results for both models. MiniMax-M3 has 41 scored results on Noometry and Qwen2.5 7B Instruct has 15.