Model comparison
GPT-4 vs Qwen2-72B
GPT-4 and Qwen2-72B score almost the same on the Noometry Index (29.1 vs 30.0), so choose on price, context window or the category you care about most.
Last verified . 25 shared benchmarks.
Alibaba (Qwen)Summary
- They share 25 benchmarks with published results for both. GPT-4 scores higher in 4 categories and Qwen2-72B in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen2-72B leads 30.2 to 10.8.
- The biggest single-benchmark swing is MATH Level 5: 23% for GPT-4 and 39.1% for Qwen2-72B.
- Qwen2-72B has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Qwen2-72B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 29.1 | 30.0 |
| Released | 2023-03-14 | 2024-06-07 |
| Weights | Proprietary | Open |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 26 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4 leads
GPT-4: 31.6 (#283), Qwen2-72B: 29.1 (#310)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| WeirdML | 12.4% | 11.3% |
| BigCodeBench Instruct | 46% | 38.5% |
| LMArena Coding | 1254 | 1196 |
| BigCodeBench Complete | 57.2% | 54% |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Qwen2-72B: 17.0 (#146)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| METR Time Horizons | 36.1% | 29.9% |
| TheAgentCompany | — | 1.1% |
Reasoning Qwen2-72B leads
GPT-4: 17.8 (#289), Qwen2-72B: 23.2 (#181)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1191 |
| Epoch Capabilities Index | 125.89 | 125.28 |
| Chess Puzzles | 4% | — |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Qwen2-72B leads
GPT-4: 10.8 (#309), Qwen2-72B: 30.2 (#236)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| LMArena Math | 1269 | 1235 |
| MATH Level 5 | 23% | 39.1% |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| GSM8K | 92% | — |
Knowledge Qwen2-72B leads
GPT-4: 18.4 (#282), Qwen2-72B: 21.2 (#275)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| GPQA Diamond | 35.7% | 40.8% |
| LMArena Expert | 1211 | 1171 |
| MMLU | 86.4% | 82.4% |
| TriviaQA | 84.8% | — |
Multilingual GPT-4 leads
GPT-4: 40.6 (#215), Qwen2-72B: 35.9 (#244)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| LMArena Non-English | 1246 | 1176 |
| LMArena Chinese | 1242 | 1240 |
| LMArena French | 1283 | 1170 |
| LMArena German | 1251 | 1151 |
| LMArena Japanese | 1209 | 1111 |
| LMArena Korean | 1184 | 1083 |
| LMArena Russian | 1251 | 1169 |
| LMArena Spanish | 1261 | 1169 |
Instruction Following GPT-4 leads
GPT-4: 65.3 (#222), Qwen2-72B: 61.7 (#241)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| LMArena Instruction Following | 1241 | 1181 |
Long Context GPT-4 leads
GPT-4: 37.7 (#212), Qwen2-72B: 36.1 (#235)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| LMArena Longer Query | 1244 | 1192 |
Writing & Preference Qwen2-72B leads
GPT-4: 34.9 (#268), Qwen2-72B: 40.8 (#241)
| Benchmark | GPT-4 | Qwen2-72B |
|---|---|---|
| LMArena Text | 1263 | 1203 |
| LMArena Creative Writing | 1244 | 1181 |
| LMArena Multi-Turn | 1257 | 1196 |
| EQ-Bench Creative Writing | 752 | — |
Frequently asked questions
Is GPT-4 better than Qwen2-72B?
GPT-4 and Qwen2-72B score almost the same on the Noometry Index (29.1 vs 30.0), so choose on price, context window or the category you care about most.
Is GPT-4 or Qwen2-72B better for coding?
GPT-4 scores higher on coding benchmarks: 31.6 versus 29.1 in the Noometry coding category.
How many benchmarks do GPT-4 and Qwen2-72B share?
25 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Qwen2-72B has 26.