Model comparison
GPT-4.1 vs Qwen3.8 27B
Qwen3.8 27B is the stronger model overall, scoring 45.3 to 35.7 on the Noometry Index.
Last verified . 22 shared benchmarks.
Alibaba (Qwen)Summary
- They share 22 benchmarks with published results for both. GPT-4.1 scores higher in 0 categories and Qwen3.8 27B in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.8 27B leads 31.5 to 11.7.
- The biggest single-benchmark swing is DTBench: 68.3% for GPT-4.1 and 88% for Qwen3.8 27B.
- Qwen3.8 27B is cheaper at $0.99 / $1.49 per million input/output tokens, against $2 / $8 for GPT-4.1.
- GPT-4.1 accepts more context: 1.05M tokens versus 262K.
- Qwen3.8 27B has downloadable open weights; the other is API-only.
Side by side
| GPT-4.1 | Qwen3.8 27B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 35.7 | 45.3 |
| Released | 2025-04-14 | 2026-08-14 |
| Weights | Proprietary | Open |
| Context window | 1.05M | 262K |
| Max output | 33K | 33K |
| Input $ / M tokens | $2 | $0.99 |
| Output $ / M tokens | $8 | $1.49 |
| Results tracked | 52 | 28 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 27B leads
GPT-4.1: 34.2 (#228), Qwen3.8 27B: 50.3 (#42)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Coding | 1391 | 1487 |
| SWE-bench Verified | 48.5% | — |
| SWE-bench Verified (bash only) | 39.6% | — |
| Aider Polyglot | 52.4% | — |
| LMArena WebDev | — | 1592 |
| SciCode | — | 46.6% |
| WeirdML | 39% | — |
| CadEval | 42% | — |
| ALE-Bench | 558.1 | — |
Agentic & Tool Use Not comparable
GPT-4.1: 34.4 (#40), Qwen3.8 27B: —
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 54% | — |
Reasoning Qwen3.8 27B leads
GPT-4.1: 11.7 (#334), Qwen3.8 27B: 31.5 (#71)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Hard Prompts | 1384 | 1459 |
| DTBench | 68.3% | 88% |
| LMCA | 25.6% | 41.4% |
| Epoch Capabilities Index | 136.79 | 149.38 |
| ARC-AGI-2 | 0.4% | — |
| SimpleBench | 27% | — |
| Kagi LLM Benchmark | 52.3% | — |
| NYT Connections (extended) | — | 54.5% |
| ARC-AGI-1 | 5.5% | — |
| CritPt | — | 5.4% |
| Chess Puzzles | 6% | — |
| EnigmaEval | 2.2% | — |
| Surface Evolver Bench | — | 45% |
| ForecastBench | 61.5 | — |
Math Qwen3.8 27B leads
GPT-4.1: 21.4 (#269), Qwen3.8 27B: 35.7 (#139)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Math | 1369 | 1453 |
| FrontierMath (Tiers 1-3) | 6% | — |
| OTIS Mock AIME 2024-2025 | 38.3% | — |
| ProofBench | — | 16% |
| Omni-MATH | 47.1% | — |
| MATH Level 5 | 83% | — |
| FrontierMath (Feb 2025 set) | 5.5% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge Qwen3.8 27B leads
GPT-4.1: 37.0 (#152), Qwen3.8 27B: 40.9 (#105)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1364 | 1482 |
| GPQA Diamond | 66.9% | — |
| Humanity's Last Exam | 5.4% | — |
| SimpleQA Verified | 31.1% | — |
| MMLU-Pro | 81.1% | — |
| Vectara Hallucination Rate | 5.6% | — |
| GPQA (HELM) | 65.9% | — |
Multimodal Qwen3.8 27B leads
GPT-4.1: 37.5 (#66), Qwen3.8 27B: 40.5 (#36)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | 1210 | 1270 |
| GeoBench | 72% | — |
Multilingual Qwen3.8 27B leads
GPT-4.1: 49.6 (#128), Qwen3.8 27B: 54.0 (#54)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1370 | 1431 |
| LMArena Chinese | 1382 | 1504 |
| LMArena French | 1383 | 1468 |
| LMArena German | 1380 | 1435 |
| LMArena Japanese | 1320 | 1392 |
| LMArena Korean | 1339 | 1389 |
| LMArena Russian | 1377 | 1421 |
| LMArena Spanish | 1375 | 1444 |
Instruction Following Qwen3.8 27B leads
GPT-4.1: 71.5 (#148), Qwen3.8 27B: 75.8 (#54)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1367 | 1436 |
| IFEval | 83.8% | — |
Long Context Qwen3.8 27B leads
GPT-4.1: 40.0 (#158), Qwen3.8 27B: 44.3 (#66)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1385 | 1448 |
| Fiction.LiveBench | 63.9% | — |
Writing & Preference Qwen3.8 27B leads
GPT-4.1: 57.8 (#120), Qwen3.8 27B: 65.9 (#41)
| Benchmark | GPT-4.1 | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1383 | 1441 |
| LMArena Creative Writing | 1363 | 1382 |
| EQ-Bench Creative Writing | 1420 | 1671 |
| LMArena Multi-Turn | 1398 | 1443 |
| WildBench | 85.4% | — |
Frequently asked questions
Is GPT-4.1 better than Qwen3.8 27B?
Qwen3.8 27B is the stronger model overall, scoring 45.3 to 35.7 on the Noometry Index.
Which is cheaper, GPT-4.1 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; GPT-4.1 lists at $2 and $8.
Is GPT-4.1 or Qwen3.8 27B better for coding?
Qwen3.8 27B scores higher on coding benchmarks: 50.3 versus 34.2 in the Noometry coding category.
Which has the bigger context window?
GPT-4.1 does, with 1.05M tokens against 262K.
How many benchmarks do GPT-4.1 and Qwen3.8 27B share?
22 benchmarks have published results for both models. GPT-4.1 has 52 scored results on Noometry and Qwen3.8 27B has 28.