Model comparison
GPT-3.5-turbo vs Qwen2.5-Coder-32B
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 23.2 on the Noometry Index.
Last verified . 21 shared benchmarks.
Alibaba (Qwen)Summary
- They share 21 benchmarks with published results for both. GPT-3.5-turbo scores higher in 1 category and Qwen2.5-Coder-32B in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen2.5-Coder-32B leads 33.3 to 6.3.
- The biggest single-benchmark swing is BigCodeBench Instruct: 39.1% for GPT-3.5-turbo and 49% for Qwen2.5-Coder-32B.
- Both cost about the same: $0.50 input and $1.50 output per million tokens.
- Qwen2.5-Coder-32B accepts more context: 33K tokens versus 16K.
- Qwen2.5-Coder-32B has downloadable open weights; the other is API-only.
Side by side
| GPT-3.5-turbo | Qwen2.5-Coder-32B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 23.2 | 33.4 |
| Released | 2023-03-01 | 2024-09-18 |
| Weights | Proprietary | Open |
| Context window | 16K | 33K |
| Max output | 4K | 29K |
| Input $ / M tokens | $0.50 | $0.66 |
| Output $ / M tokens | $1.50 | $1 |
| Results tracked | 44 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-3.5-turbo leads
GPT-3.5-turbo: 23.9 (#331), Qwen2.5-Coder-32B: 22.6 (#333)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| BigCodeBench Instruct | 39.1% | 49% |
| LMArena Coding | 1136 | 1276 |
| BigCodeBench Complete | 50.6% | 58% |
| HumanEval+ | 70.7% | 87.2% |
| MBPP+ | 69.7% | 77% |
| SWE-bench Verified (bash only) | — | 9% |
| Aider Polyglot | — | 16.4% |
| WeirdML | 3.5% | — |
| LiveBench Coding | — | 56.9% |
Agentic & Tool Use Not comparable
GPT-3.5-turbo: —, Qwen2.5-Coder-32B: —
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| METR Time Horizons | 21.5% | — |
Reasoning Qwen2.5-Coder-32B leads
GPT-3.5-turbo: 13.8 (#332), Qwen2.5-Coder-32B: 21.2 (#225)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Hard Prompts | 1108 | 1251 |
| Epoch Capabilities Index | 118.55 | 119.49 |
| WinoGrande | 81.6% | 80.8% |
| Chess Puzzles | 0% | — |
| LiveBench Reasoning | — | 42.1% |
| Mystery Game Puzzles | 3% | — |
| DTBench | 48.5% | — |
| LiveBench Data Analysis | — | 49.9% |
| LMCA | 9.7% | — |
| Adversarial NLI | 58.1% | — |
| BIG-Bench Hard | 61.6% | — |
| CommonsenseQA 2.0 | 57% | — |
| ForecastBench | 50.4 | — |
| HellaSwag | — | 83% |
| LiveBench | — | 46.2% |
Math Qwen2.5-Coder-32B leads
GPT-3.5-turbo: 6.3 (#327), Qwen2.5-Coder-32B: 33.3 (#204)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Math | 1142 | 1251 |
| GSM8K | 57.8% | 93% |
| FrontierMath (Tiers 1-3) | 0% | — |
| OTIS Mock AIME 2024-2025 | 2.2% | — |
| LiveBench Math | — | 46.6% |
| MATH Level 5 | 15.9% | — |
Knowledge Qwen2.5-Coder-32B leads
GPT-3.5-turbo: 10.0 (#303), Qwen2.5-Coder-32B: 33.4 (#203)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Expert | 1070 | 1221 |
| ARC (AI2) Challenge | 87.4% | 70.5% |
| MMLU | 71.4% | 79.1% |
| GPQA Diamond | 28% | — |
| BoolQ | 87% | — |
| OpenBookQA | 86% | — |
| TriviaQA | 85.8% | — |
Multilingual Qwen2.5-Coder-32B leads
GPT-3.5-turbo: 31.5 (#258), Qwen2.5-Coder-32B: 37.8 (#235)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Non-English | 1108 | 1205 |
| LMArena Chinese | 1075 | 1222 |
| LMArena Russian | 1123 | 1228 |
| LMArena French | 1118 | — |
| LMArena German | 1090 | — |
| LMArena Japanese | 1043 | — |
| LMArena Korean | 1019 | — |
| LMArena Spanish | 1121 | — |
Instruction Following Qwen2.5-Coder-32B leads
GPT-3.5-turbo: 57.9 (#262), Qwen2.5-Coder-32B: 61.4 (#245)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Instruction Following | 1119 | 1223 |
| LiveBench Instruction Following | — | 58.7% |
Long Context Qwen2.5-Coder-32B leads
GPT-3.5-turbo: 34.0 (#254), Qwen2.5-Coder-32B: 38.0 (#208)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Longer Query | 1121 | 1251 |
Writing & Preference Qwen2.5-Coder-32B leads
GPT-3.5-turbo: 25.3 (#305), Qwen2.5-Coder-32B: 41.6 (#240)
| Benchmark | GPT-3.5-turbo | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Text | 1125 | 1230 |
| LMArena Creative Writing | 1092 | 1174 |
| LMArena Multi-Turn | 1117 | 1222 |
| EQ-Bench Creative Writing | 451 | — |
| LiveBench Language | — | 23.3% |
Frequently asked questions
Is GPT-3.5-turbo better than Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 23.2 on the Noometry Index.
Which is cheaper, GPT-3.5-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-3.5-turbo lists at $0.50 and $1.50.
Is GPT-3.5-turbo or Qwen2.5-Coder-32B better for coding?
GPT-3.5-turbo scores higher on coding benchmarks: 23.9 versus 22.6 in the Noometry coding category.
Which has the bigger context window?
Qwen2.5-Coder-32B does, with 33K tokens against 16K.
How many benchmarks do GPT-3.5-turbo and Qwen2.5-Coder-32B share?
21 benchmarks have published results for both models. GPT-3.5-turbo has 44 scored results on Noometry and Qwen2.5-Coder-32B has 31.