# Claude Sonnet 4 vs Qwen2.5 72B Instruct

> Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 31.5 on the Noometry Index. Qwen2.5 72B Instruct costs 2.4× less per token, which makes it the better buy when Claude Sonnet 4's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/claude-sonnet-4-vs-qwen2-5-72b-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 33

## Summary

- They share 33 benchmarks with published results for both. Claude Sonnet 4 scores higher in 8 categories and Qwen2.5 72B Instruct in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Sonnet 4 leads 41.5 to 18.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 71.1% for Claude Sonnet 4 and 8.1% for Qwen2.5 72B Instruct.
- Qwen2.5 72B Instruct is cheaper at $1.40 / $5.60 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- Claude Sonnet 4 accepts more context: 200K tokens versus 131K.
- Qwen2.5 72B Instruct has downloadable open weights; the other is API-only.

## Snapshot

| | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| Provider | Anthropic | Alibaba (Qwen) |
| Noometry Index | 40.6 | 31.5 |
| Rank | 139 | 259 |
| Context | 200K | 131K |
| Input $/M | $3 | $1.40 |
| Output $/M | $15 | $5.60 |
| Weights | Proprietary | Open |

## Coding

- Claude Sonnet 4: 44.0 (#70)
- Qwen2.5 72B Instruct: 32.8 (#256)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 46.1% | 16% |
| LMArena Coding | 1414 | 1293 |
| SWE-bench Verified (bash only) | 64.9% | — |
| Aider Polyglot | 61.3% | — |
| SciCode | 40% | — |
| GSO | 4.9% | — |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 655.35 | — |

## Agentic & Tool Use

- Claude Sonnet 4: 38.3 (#29)
- Qwen2.5 72B Instruct: 22.0 (#124)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| TheAgentCompany | 33.1% | 5.7% |
| METR Time Horizons | 62% | 35.8% |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| BALROG | — | 16.2% |

## Reasoning

- Claude Sonnet 4: 23.0 (#169)
- Qwen2.5 72B Instruct: 21.4 (#197)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1372 | 1271 |
| DTBench | 77.1% | 62.9% |
| LMCA | 29% | 13.4% |
| Epoch Capabilities Index | 141.69 | 129.01 |
| ForecastBench | 60.2 | 57.5 |
| ARC-AGI-2 | 5.9% | — |
| SimpleBench | 45.5% | — |
| Kagi LLM Benchmark | 73% | — |
| ARC-AGI-1 | 40% | — |
| CritPt | 0.3% | — |
| EnigmaEval | 3.1% | — |
| BIG-Bench Hard | — | 79.8% |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |

## Math

- Claude Sonnet 4: 41.5 (#71)
- Qwen2.5 72B Instruct: 18.1 (#281)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 8.1% |
| Omni-MATH | 60.2% | 33% |
| LMArena Math | 1375 | 1283 |
| MATH Level 5 | 84.4% | 63.2% |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |

## Knowledge

- Claude Sonnet 4: 41.6 (#99)
- Qwen2.5 72B Instruct: 26.9 (#247)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 79.2% | 49.1% |
| MMLU-Pro | 84.3% | 63.1% |
| Confabulations | 13.2% | 19.1% |
| GPQA (HELM) | 70.6% | 42.6% |
| LMArena Expert | 1371 | 1245 |
| Humanity's Last Exam | 7.8% | — |
| Vectara Hallucination Rate | 10.3% | — |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |

## Multimodal

- Claude Sonnet 4: 25.6 (#116)
- Qwen2.5 72B Instruct: —

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Vision | 1191 | — |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |

## Multilingual

- Claude Sonnet 4: 46.9 (#151)
- Qwen2.5 72B Instruct: 41.2 (#208)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1333 | 1252 |
| LMArena Chinese | 1351 | 1272 |
| LMArena French | 1364 | 1279 |
| LMArena German | 1331 | 1234 |
| LMArena Japanese | 1301 | 1180 |
| LMArena Korean | 1293 | 1188 |
| LMArena Russian | 1355 | 1264 |
| LMArena Spanish | 1356 | 1255 |

## Instruction Following

- Claude Sonnet 4: 71.8 (#141)
- Qwen2.5 72B Instruct: 65.7 (#216)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | 84% | 80.6% |
| LMArena Instruction Following | 1375 | 1254 |

## Long Context

- Claude Sonnet 4: 33.6 (#254)
- Qwen2.5 72B Instruct: 38.9 (#183)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1398 | 1282 |
| Fiction.LiveBench | 46.9% | — |

## Writing & Preference

- Claude Sonnet 4: 57.3 (#127)
- Qwen2.5 72B Instruct: 46.9 (#210)

| Benchmark | Claude Sonnet 4 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1351 | 1269 |
| LMArena Creative Writing | 1345 | 1221 |
| WildBench | 83.8% | 80.2% |
| LMArena Multi-Turn | 1376 | 1272 |
| Short-Story Creative Writing | 81.4% | — |
| EQ-Bench Creative Writing | 1483 | — |

## FAQ

### Is Claude Sonnet 4 better than Qwen2.5 72B Instruct?

Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 31.5 on the Noometry Index. Qwen2.5 72B Instruct costs 2.4× less per token, which makes it the better buy when Claude Sonnet 4's lead doesn't matter for your workload.

### Which is cheaper, Claude Sonnet 4 or Qwen2.5 72B Instruct?

Qwen2.5 72B Instruct is cheaper. It lists at $1.40 per million input tokens and $5.60 per million output tokens; Claude Sonnet 4 lists at $3 and $15.

### Is Claude Sonnet 4 or Qwen2.5 72B Instruct better for coding?

Claude Sonnet 4 scores higher on coding benchmarks: 44.0 versus 32.8 in the Noometry coding category.

### Which has the bigger context window?

Claude Sonnet 4 does, with 200K tokens against 131K.

### How many benchmarks do Claude Sonnet 4 and Qwen2.5 72B Instruct share?

33 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and Qwen2.5 72B Instruct has 43.