Model comparison
GPT-4 Turbo vs Qwen2.5-Coder-32B
Qwen2.5-Coder-32B is the stronger model overall, scoring 32.9 to 30.3 on the Noometry Index.
Last verified . 18 shared benchmarks.
Alibaba (Qwen)Summary
- They share 18 benchmarks with published results for both. GPT-4 Turbo scores higher in 5 categories and Qwen2.5-Coder-32B in 3 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen2.5-Coder-32B leads 30.8 to 8.6.
- Qwen2.5-Coder-32B is cheaper at $0.66 / $1 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- GPT-4 Turbo accepts more context: 128K tokens versus 33K.
- Qwen2.5-Coder-32B has downloadable open weights; the other is API-only.
Side by side
| GPT-4 Turbo | Qwen2.5-Coder-32B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 30.3 | 32.9 |
| Released | 2023-11-06 | 2024-09-18 |
| Weights | Proprietary | Open |
| Context window | 128K | 33K |
| Max output | 4K | 29K |
| Input $ / M tokens | $10 | $0.66 |
| Output $ / M tokens | $30 | $1 |
| Results tracked | 36 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4 Turbo leads
GPT-4 Turbo: 33.4 (#247), Qwen2.5-Coder-32B: 22.4 (#328)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| BigCodeBench Instruct | 48.2% | 49% |
| LMArena Coding | 1268 | 1276 |
| BigCodeBench Complete | 58.2% | 58% |
| HumanEval+ | 86.6% | 87.2% |
| MBPP+ | 73.3% | 77% |
| SWE-bench Verified (bash only) | — | 9% |
| Aider Polyglot | — | 16.4% |
| WeirdML | 18% | — |
| LiveBench Coding | — | 56.9% |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, Qwen2.5-Coder-32B: —
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| METR Time Horizons | 36.7% | — |
Reasoning Qwen2.5-Coder-32B leads
GPT-4 Turbo: 14.9 (#311), Qwen2.5-Coder-32B: 20.4 (#220)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Hard Prompts | 1251 | 1251 |
| Epoch Capabilities Index | 127.25 | 119.41 |
| SimpleBench | 25.1% | — |
| Chess Puzzles | 6% | — |
| LiveBench Reasoning | — | 42.1% |
| DTBench | 61.6% | — |
| LiveBench Data Analysis | — | 49.9% |
| LMCA | 9.8% | — |
| ForecastBench | 59.4 | — |
| HellaSwag | — | 83% |
| LiveBench | — | 46.2% |
| WinoGrande | — | 80.8% |
Math Qwen2.5-Coder-32B leads
GPT-4 Turbo: 8.6 (#316), Qwen2.5-Coder-32B: 30.8 (#203)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Math | 1272 | 1251 |
| FrontierMath (Tiers 1-3) | 0.7% | — |
| OTIS Mock AIME 2024-2025 | 6.7% | — |
| LiveBench Math | — | 46.6% |
| MATH Level 5 | 46.7% | — |
| GSM8K | — | 93% |
Knowledge Qwen2.5-Coder-32B leads
GPT-4 Turbo: 24.3 (#263), Qwen2.5-Coder-32B: 32.9 (#200)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Expert | 1223 | 1221 |
| MMLU | 81.3% | 79.1% |
| GPQA Diamond | 46.6% | — |
| Confabulations | 28.4% | — |
| ARC (AI2) Challenge | — | 70.5% |
Multimodal Not comparable
GPT-4 Turbo: 30.0 (#104), Qwen2.5-Coder-32B: —
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Vision | 1090 | — |
Multilingual GPT-4 Turbo leads
GPT-4 Turbo: 40.7 (#211), Qwen2.5-Coder-32B: 38.0 (#230)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Non-English | 1245 | 1205 |
| LMArena Chinese | 1242 | 1222 |
| LMArena Russian | 1259 | 1228 |
| LMArena French | 1275 | — |
| LMArena German | 1259 | — |
| LMArena Japanese | 1194 | — |
| LMArena Korean | 1187 | — |
| LMArena Spanish | 1260 | — |
Instruction Following GPT-4 Turbo leads
GPT-4 Turbo: 66.0 (#211), Qwen2.5-Coder-32B: 61.6 (#240)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Instruction Following | 1250 | 1223 |
| LiveBench Instruction Following | — | 58.7% |
Long Context Too close to call
GPT-4 Turbo: 38.0 (#200), Qwen2.5-Coder-32B: 37.9 (#204)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Longer Query | 1254 | 1251 |
Writing & Preference GPT-4 Turbo leads
GPT-4 Turbo: 47.9 (#201), Qwen2.5-Coder-32B: 41.8 (#235)
| Benchmark | GPT-4 Turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Text | 1272 | 1230 |
| LMArena Creative Writing | 1269 | 1174 |
| LMArena Multi-Turn | 1267 | 1222 |
| LiveBench Language | — | 23.3% |
Frequently asked questions
Is GPT-4 Turbo better than Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is the stronger model overall, scoring 32.9 to 30.3 on the Noometry Index.
Which is cheaper, GPT-4 Turbo or Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is cheaper. It lists at $0.66 per million input tokens and $1 per million output tokens; GPT-4 Turbo lists at $10 and $30.
Is GPT-4 Turbo or Qwen2.5-Coder-32B better for coding?
GPT-4 Turbo scores higher on coding benchmarks: 33.4 versus 22.4 in the Noometry coding category.
Which has the bigger context window?
GPT-4 Turbo does, with 128K tokens against 33K.
How many benchmarks do GPT-4 Turbo and Qwen2.5-Coder-32B share?
18 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and Qwen2.5-Coder-32B has 31.