# Qwen2-72B vs Step 3

> Step 3 is the stronger model overall, scoring 39.9 to 29.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/qwen2-72b-vs-step-3
- Last updated: 2026-10-10
- Shared benchmarks: 15

## Summary

- They share 15 benchmarks with published results for both. Qwen2-72B scores higher in 0 categories and Step 3 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Step 3 leads 36.2 to 21.2.

## Snapshot

| | Qwen2-72B | Step 3 |
|---|---|---|
| Provider | Alibaba (Qwen) | StepFun |
| Noometry Index | 29.5 | 39.9 |
| Rank | 294 | 145 |
| Context | — | — |
| Input $/M | — | — |
| Output $/M | — | — |
| Weights | Open | Open |

## Coding

- Qwen2-72B: 28.7 (#303)
- Step 3: 39.5 (#144)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Coding | 1196 | 1367 |
| WeirdML | 11.3% | — |
| BigCodeBench Instruct | 38.5% | — |
| BigCodeBench Complete | 54% | — |

## Agentic & Tool Use

- Qwen2-72B: 16.9 (#138)
- Step 3: —

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| TheAgentCompany | 1.1% | — |
| METR Time Horizons | 29.9% | — |

## Reasoning

- Qwen2-72B: 22.2 (#184)
- Step 3: 27.6 (#102)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Hard Prompts | 1191 | 1355 |
| Kagi LLM Benchmark | — | 62.3% |
| Epoch Capabilities Index | 125.27 | — |

## Math

- Qwen2-72B: 27.9 (#232)
- Step 3: 34.9 (#148)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Math | 1235 | 1366 |
| MATH Level 5 | 39.1% | — |

## Knowledge

- Qwen2-72B: 21.2 (#270)
- Step 3: 36.2 (#161)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Expert | 1171 | 1333 |
| GPQA Diamond | 40.8% | — |
| MMLU | 82.4% | — |

## Multimodal

- Qwen2-72B: —
- Step 3: 34.9 (#81)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Vision | — | 1176 |

## Multilingual

- Qwen2-72B: 36.0 (#239)
- Step 3: 46.5 (#154)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Non-English | 1176 | 1327 |
| LMArena Chinese | 1240 | 1396 |
| LMArena German | 1151 | 1371 |
| LMArena Korean | 1083 | 1271 |
| LMArena Russian | 1169 | 1331 |
| LMArena Spanish | 1168 | 1372 |
| LMArena French | 1169 | — |
| LMArena Japanese | 1111 | — |

## Instruction Following

- Qwen2-72B: 62.0 (#236)
- Step 3: 70.6 (#159)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1181 | 1332 |

## Long Context

- Qwen2-72B: 36.1 (#230)
- Step 3: 40.3 (#152)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Longer Query | 1193 | 1326 |

## Writing & Preference

- Qwen2-72B: 41.0 (#236)
- Step 3: 54.5 (#146)

| Benchmark | Qwen2-72B | Step 3 |
|---|---|---|
| LMArena Text | 1203 | 1351 |
| LMArena Creative Writing | 1181 | 1322 |
| LMArena Multi-Turn | 1195 | 1340 |

## FAQ

### Is Qwen2-72B better than Step 3?

Step 3 is the stronger model overall, scoring 39.9 to 29.5 on the Noometry Index.

### Is Qwen2-72B or Step 3 better for coding?

Step 3 scores higher on coding benchmarks: 39.5 versus 28.7 in the Noometry coding category.

### How many benchmarks do Qwen2-72B and Step 3 share?

15 benchmarks have published results for both models. Qwen2-72B has 26 scored results on Noometry and Step 3 has 17.