Model comparison
GPT-4 vs Qwen1.5-32B
Qwen1.5-32B is the stronger model overall, scoring 30.0 to 28.9 on the Noometry Index.
Last verified . 21 shared benchmarks.
Alibaba (Qwen)Summary
- They share 21 benchmarks with published results for both. GPT-4 scores higher in 6 categories and Qwen1.5-32B in 2 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen1.5-32B leads 30.5 to 10.1.
- The biggest single-benchmark swing is BigCodeBench Complete: 57.2% for GPT-4 and 42% for Qwen1.5-32B.
- Qwen1.5-32B has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Qwen1.5-32B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 28.9 | 30.0 |
| Released | 2023-03-14 | 2024-02-04 |
| Weights | Proprietary | Open |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 21 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
GPT-4: 31.2 (#277), Qwen1.5-32B: 31.1 (#278)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| BigCodeBench Instruct | 46% | 32.3% |
| LMArena Coding | 1254 | 1155 |
| BigCodeBench Complete | 57.2% | 42% |
| WeirdML | 12.4% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Qwen1.5-32B: —
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning Qwen1.5-32B leads
GPT-4: 17.3 (#285), Qwen1.5-32B: 20.9 (#210)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1131 |
| Chess Puzzles | 4% | — |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| Epoch Capabilities Index | 125.88 | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Qwen1.5-32B leads
GPT-4: 10.1 (#304), Qwen1.5-32B: 30.5 (#207)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| LMArena Math | 1269 | 1155 |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| MATH Level 5 | 23% | — |
| GSM8K | 92% | — |
Knowledge GPT-4 leads
GPT-4: 18.4 (#277), Qwen1.5-32B: 13.5 (#291)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| GPQA Diamond | 35.7% | 30.7% |
| LMArena Expert | 1211 | 1126 |
| MMLU | 86.4% | 74.4% |
| TriviaQA | 84.8% | — |
Multilingual GPT-4 leads
GPT-4: 40.8 (#210), Qwen1.5-32B: 31.6 (#254)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| LMArena Non-English | 1246 | 1106 |
| LMArena Chinese | 1242 | 1176 |
| LMArena French | 1282 | 1100 |
| LMArena German | 1251 | 1058 |
| LMArena Japanese | 1209 | 1027 |
| LMArena Korean | 1184 | 1007 |
| LMArena Russian | 1251 | 1073 |
| LMArena Spanish | 1261 | 1089 |
Instruction Following GPT-4 leads
GPT-4: 65.5 (#217), Qwen1.5-32B: 58.0 (#260)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| LMArena Instruction Following | 1241 | 1116 |
Long Context GPT-4 leads
GPT-4: 37.7 (#207), Qwen1.5-32B: 34.7 (#241)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| LMArena Longer Query | 1244 | 1146 |
Writing & Preference Too close to call
GPT-4: 35.0 (#263), Qwen1.5-32B: 34.4 (#266)
| Benchmark | GPT-4 | Qwen1.5-32B |
|---|---|---|
| LMArena Text | 1263 | 1137 |
| LMArena Creative Writing | 1244 | 1083 |
| LMArena Multi-Turn | 1257 | 1140 |
| EQ-Bench Creative Writing | 752 | — |
Frequently asked questions
Is GPT-4 better than Qwen1.5-32B?
Qwen1.5-32B is the stronger model overall, scoring 30.0 to 28.9 on the Noometry Index.
Is GPT-4 or Qwen1.5-32B better for coding?
They score almost the same on coding (31.2 vs 31.1); test both on your own repository before choosing.
How many benchmarks do GPT-4 and Qwen1.5-32B share?
21 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Qwen1.5-32B has 21.