# Qwen3.5 27B vs Step 3

> Qwen3.5 27B is the stronger model overall, scoring 41.9 to 40.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/qwen3-5-27b-vs-step-3
- Last updated: 2026-10-10
- Shared benchmarks: 16

## Summary

- They share 16 benchmarks with published results for both. Qwen3.5 27B scores higher in 7 categories and Step 3 in 2 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Qwen3.5 27B leads 59.3 to 54.3.

## Snapshot

| | Qwen3.5 27B | Step 3 |
|---|---|---|
| Provider | Alibaba (Qwen) | StepFun |
| Noometry Index | 41.9 | 40.5 |
| Rank | 127 | 149 |
| Context | 262K | — |
| Input $/M | $0.30 | — |
| Output $/M | $2.40 | — |
| Weights | Open | Open |

## Coding

- Qwen3.5 27B: 38.9 (#168)
- Step 3: 40.1 (#147)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Coding | 1427 | 1367 |
| LMArena WebDev | 1358 | — |
| WeirdML | 39.5% | — |
| ALE-Bench | 349.45 | — |

## Agentic & Tool Use

- Qwen3.5 27B: —
- Step 3: —

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| Vending-Bench 2 | 201.98 | — |

## Reasoning

- Qwen3.5 27B: 27.5 (#117)
- Step 3: 28.4 (#105)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Hard Prompts | 1414 | 1355 |
| Kagi LLM Benchmark | — | 62.3% |
| NYT Connections (extended) | 47.9% | — |
| Thematic Generalization | 45.5% | — |
| DTBench | 82.4% | — |
| LMCA | 34% | — |

## Math

- Qwen3.5 27B: 38.8 (#127)
- Step 3: 37.6 (#148)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Math | 1429 | 1366 |
| MathArena Final-Answer Competitions | 56.7% | — |

## Knowledge

- Qwen3.5 27B: 38.0 (#150)
- Step 3: 36.8 (#164)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Expert | 1428 | 1333 |
| Vectara Hallucination Rate | 12.1% | — |

## Multimodal

- Qwen3.5 27B: 39.4 (#59)
- Step 3: 35.5 (#86)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Vision | 1241 | 1177 |

## Multilingual

- Qwen3.5 27B: 50.8 (#115)
- Step 3: 46.3 (#159)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Non-English | 1390 | 1327 |
| LMArena Chinese | 1478 | 1397 |
| LMArena German | 1393 | 1371 |
| LMArena Korean | 1358 | 1269 |
| LMArena Russian | 1390 | 1331 |
| LMArena Spanish | 1407 | 1371 |
| LMArena French | 1410 | — |
| LMArena Japanese | 1345 | — |

## Instruction Following

- Qwen3.5 27B: 73.5 (#119)
- Step 3: 70.4 (#164)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1393 | 1332 |

## Long Context

- Qwen3.5 27B: 43.1 (#106)
- Step 3: 40.3 (#157)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Longer Query | 1413 | 1326 |

## Writing & Preference

- Qwen3.5 27B: 59.3 (#111)
- Step 3: 54.3 (#151)

| Benchmark | Qwen3.5 27B | Step 3 |
|---|---|---|
| LMArena Text | 1409 | 1350 |
| LMArena Creative Writing | 1362 | 1321 |
| LMArena Multi-Turn | 1410 | 1341 |

## FAQ

### Is Qwen3.5 27B better than Step 3?

Qwen3.5 27B is the stronger model overall, scoring 41.9 to 40.5 on the Noometry Index.

### Is Qwen3.5 27B or Step 3 better for coding?

Step 3 scores higher on coding benchmarks: 40.1 versus 38.9 in the Noometry coding category.

### How many benchmarks do Qwen3.5 27B and Step 3 share?

16 benchmarks have published results for both models. Qwen3.5 27B has 28 scored results on Noometry and Step 3 has 17.