Model comparison
Claude Sonnet 4 vs Qwen3.8 27B
Qwen3.8 27B is the stronger model overall, scoring 45.3 to 40.6 on the Noometry Index.
Last verified . 24 shared benchmarks.
Alibaba (Qwen)Summary
- They share 24 benchmarks with published results for both. Claude Sonnet 4 scores higher in 2 categories and Qwen3.8 27B in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in multimodal, where Qwen3.8 27B leads 40.5 to 25.6.
- The biggest single-benchmark swing is LMCA: 29% for Claude Sonnet 4 and 41.4% for Qwen3.8 27B.
- Qwen3.8 27B is cheaper at $0.99 / $1.49 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- Qwen3.8 27B accepts more context: 262K tokens versus 200K.
- Qwen3.8 27B has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 4 | Qwen3.8 27B | |
|---|---|---|
| Provider | Anthropic | Alibaba (Qwen) |
| Noometry Index | 40.6 | 45.3 |
| Released | 2025-05-22 | 2026-08-14 |
| Weights | Proprietary | Open |
| Context window | 200K | 262K |
| Max output | 64K | 33K |
| Input $ / M tokens | $3 | $0.99 |
| Output $ / M tokens | $15 | $1.49 |
| Results tracked | 58 | 28 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 27B leads
Claude Sonnet 4: 44.0 (#70), Qwen3.8 27B: 50.3 (#42)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| SciCode | 40% | 46.6% |
| LMArena Coding | 1414 | 1487 |
| SWE-bench Verified (bash only) | 64.9% | — |
| Aider Polyglot | 61.3% | — |
| LMArena WebDev | — | 1592 |
| GSO | 4.9% | — |
| WeirdML | 46.1% | — |
| ALE-Bench | 655.35 | — |
Agentic & Tool Use Not comparable
Claude Sonnet 4: 38.3 (#29), Qwen3.8 27B: —
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| METR Time Horizons | 62% | — |
Reasoning Qwen3.8 27B leads
Claude Sonnet 4: 23.0 (#169), Qwen3.8 27B: 31.5 (#71)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| CritPt | 0.3% | 5.4% |
| LMArena Hard Prompts | 1372 | 1459 |
| DTBench | 77.1% | 88% |
| LMCA | 29% | 41.4% |
| Epoch Capabilities Index | 141.69 | 149.38 |
| ARC-AGI-2 | 5.9% | — |
| SimpleBench | 45.5% | — |
| Kagi LLM Benchmark | 73% | — |
| NYT Connections (extended) | — | 54.5% |
| ARC-AGI-1 | 40% | — |
| EnigmaEval | 3.1% | — |
| Surface Evolver Bench | — | 45% |
| ForecastBench | 60.2 | — |
Math Claude Sonnet 4 leads
Claude Sonnet 4: 41.5 (#71), Qwen3.8 27B: 35.7 (#139)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| LMArena Math | 1375 | 1453 |
| OTIS Mock AIME 2024-2025 | 71.1% | — |
| ProofBench | — | 16% |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge Too close to call
Claude Sonnet 4: 41.6 (#99), Qwen3.8 27B: 40.9 (#105)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1371 | 1482 |
| GPQA Diamond | 79.2% | — |
| Humanity's Last Exam | 7.8% | — |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| Vectara Hallucination Rate | 10.3% | — |
| GPQA (HELM) | 70.6% | — |
Multimodal Qwen3.8 27B leads
Claude Sonnet 4: 25.6 (#116), Qwen3.8 27B: 40.5 (#36)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | 1191 | 1270 |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |
Multilingual Qwen3.8 27B leads
Claude Sonnet 4: 46.9 (#151), Qwen3.8 27B: 54.0 (#54)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1333 | 1431 |
| LMArena Chinese | 1351 | 1504 |
| LMArena French | 1364 | 1468 |
| LMArena German | 1331 | 1435 |
| LMArena Japanese | 1301 | 1392 |
| LMArena Korean | 1293 | 1389 |
| LMArena Russian | 1355 | 1421 |
| LMArena Spanish | 1356 | 1444 |
Instruction Following Qwen3.8 27B leads
Claude Sonnet 4: 71.8 (#141), Qwen3.8 27B: 75.8 (#54)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1375 | 1436 |
| IFEval | 84% | — |
Long Context Qwen3.8 27B leads
Claude Sonnet 4: 33.6 (#254), Qwen3.8 27B: 44.3 (#66)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1398 | 1448 |
| Fiction.LiveBench | 46.9% | — |
Writing & Preference Qwen3.8 27B leads
Claude Sonnet 4: 57.3 (#127), Qwen3.8 27B: 65.9 (#41)
| Benchmark | Claude Sonnet 4 | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1351 | 1441 |
| LMArena Creative Writing | 1345 | 1382 |
| EQ-Bench Creative Writing | 1483 | 1671 |
| LMArena Multi-Turn | 1376 | 1443 |
| Short-Story Creative Writing | 81.4% | — |
| WildBench | 83.8% | — |
Frequently asked questions
Is Claude Sonnet 4 better than Qwen3.8 27B?
Qwen3.8 27B is the stronger model overall, scoring 45.3 to 40.6 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 or Qwen3.8 27B?
Qwen3.8 27B is cheaper. It lists at $0.99 per million input tokens and $1.49 per million output tokens; Claude Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or Qwen3.8 27B better for coding?
Qwen3.8 27B scores higher on coding benchmarks: 50.3 versus 44.0 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 27B does, with 262K tokens against 200K.
How many benchmarks do Claude Sonnet 4 and Qwen3.8 27B share?
24 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and Qwen3.8 27B has 28.