# Qwen2.5-Max vs Qwen3-4B

> Qwen2.5-Max is the stronger model overall, scoring 40.7 to 31.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/qwen2-5-max-vs-qwen3-4b
- Last updated: 2026-10-12
- Shared benchmarks: 0

## Summary

- The widest gap is in math, where Qwen2.5-Max leads 36.9 to 29.7.
- Qwen3-4B has downloadable open weights; the other is API-only.

## Snapshot

| | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| Provider | Alibaba (Qwen) | Alibaba (Qwen) |
| Noometry Index | 40.7 | 31.9 |
| Rank | 146 | 264 |
| Context | — | — |
| Input $/M | — | — |
| Output $/M | — | — |
| Weights | Proprietary | Open |

## Coding

- Qwen2.5-Max: 41.8 (#117)
- Qwen3-4B: —

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| LiveBench Coding | 64.4% | — |
| LMArena Coding | 1359 | — |

## Agentic & Tool Use

- Qwen2.5-Max: —
- Qwen3-4B: 27.6 (#100)

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 35.7% |

## Reasoning

- Qwen2.5-Max: 25.6 (#147)
- Qwen3-4B: 19.2 (#268)

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| Chess Puzzles | — | 4% |
| LiveBench Reasoning | 51.4% | — |
| LMArena Hard Prompts | 1360 | — |
| LiveBench Data Analysis | 67.9% | — |
| Epoch Capabilities Index | 132.53 | — |
| LiveBench | 62.3% | — |

## Math

- Qwen2.5-Max: 36.9 (#162)
- Qwen3-4B: 29.7 (#240)

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| MathArena Final-Answer Competitions | — | 38.5% |
| OTIS Mock AIME 2024-2025 | — | 52.2% |
| LiveBench Math | 58.4% | — |
| LMArena Math | 1369 | — |

## Knowledge

- Qwen2.5-Max: 35.3 (#186)
- Qwen3-4B: 33.0 (#208)

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| GPQA Diamond | — | 52.3% |
| Confabulations | 21.8% | — |
| Vectara Hallucination Rate | — | 5.7% |
| LMArena Expert | 1337 | — |

## Multilingual

- Qwen2.5-Max: 48.1 (#146)
- Qwen3-4B: —

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| LMArena Non-English | 1352 | — |
| LMArena Chinese | 1382 | — |
| LMArena French | 1396 | — |
| LMArena German | 1350 | — |
| LMArena Japanese | 1300 | — |
| LMArena Korean | 1304 | — |
| LMArena Russian | 1353 | — |
| LMArena Spanish | 1377 | — |

## Instruction Following

- Qwen2.5-Max: 71.3 (#152)
- Qwen3-4B: —

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| LiveBench Instruction Following | 75.3% | — |
| LMArena Instruction Following | 1335 | — |

## Long Context

- Qwen2.5-Max: 41.4 (#142)
- Qwen3-4B: —

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| LMArena Longer Query | 1358 | — |

## Writing & Preference

- Qwen2.5-Max: 55.4 (#146)
- Qwen3-4B: —

| Benchmark | Qwen2.5-Max | Qwen3-4B |
|---|---|---|
| LMArena Text | 1367 | — |
| LMArena Creative Writing | 1339 | — |
| Short-Story Creative Writing | 72.9% | — |
| LMArena Multi-Turn | 1364 | — |
| LiveBench Language | 56.3% | — |

## FAQ

### Is Qwen2.5-Max better than Qwen3-4B?

Qwen2.5-Max is the stronger model overall, scoring 40.7 to 31.9 on the Noometry Index.

### How many benchmarks do Qwen2.5-Max and Qwen3-4B share?

0 benchmarks have published results for both models. Qwen2.5-Max has 27 scored results on Noometry and Qwen3-4B has 6.