Model comparison
DeepSeek-R1-Distill-Qwen-32B vs GPT-3.5-turbo
DeepSeek-R1-Distill-Qwen-32B is the stronger model overall, scoring 35.5 to 23.2 on the Noometry Index.
Last verified . 6 shared benchmarks.
DeepSeekSummary
- They share 6 benchmarks with published results for both. DeepSeek-R1-Distill-Qwen-32B scores higher in 6 categories and GPT-3.5-turbo in 0 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-R1-Distill-Qwen-32B leads 34.5 to 6.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 55.6% for DeepSeek-R1-Distill-Qwen-32B and 2.2% for GPT-3.5-turbo.
- DeepSeek-R1-Distill-Qwen-32B has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 35.5 | 23.2 |
| Released | 2025-01-20 | 2023-03-01 |
| Weights | Open | Proprietary |
| Context window | — | 16K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.50 |
| Output $ / M tokens | — | $1.50 |
| Results tracked | 14 | 44 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1-Distill-Qwen-32B leads
DeepSeek-R1-Distill-Qwen-32B: 36.1 (#212), GPT-3.5-turbo: 23.9 (#331)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| BigCodeBench Instruct | 43.9% | 39.1% |
| BigCodeBench Complete | 54.9% | 50.6% |
| WeirdML | — | 3.5% |
| LiveBench Coding | 33.7% | — |
| LMArena Coding | — | 1136 |
| HumanEval+ | — | 70.7% |
| MBPP+ | — | 69.7% |
Agentic & Tool Use Not comparable
DeepSeek-R1-Distill-Qwen-32B: 28.1 (#94), GPT-3.5-turbo: —
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| BALROG | 19.5% | — |
| METR Time Horizons | — | 21.5% |
Reasoning DeepSeek-R1-Distill-Qwen-32B leads
DeepSeek-R1-Distill-Qwen-32B: 18.2 (#284), GPT-3.5-turbo: 13.8 (#332)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| Chess Puzzles | 1% | 0% |
| Epoch Capabilities Index | 137.44 | 118.55 |
| LiveBench Reasoning | 52.3% | — |
| LMArena Hard Prompts | — | 1108 |
| Mystery Game Puzzles | — | 3% |
| DTBench | — | 48.5% |
| LiveBench Data Analysis | 45.4% | — |
| LMCA | — | 9.7% |
| Adversarial NLI | — | 58.1% |
| BIG-Bench Hard | — | 61.6% |
| CommonsenseQA 2.0 | — | 57% |
| ForecastBench | — | 50.4 |
| LiveBench | 45.5% | — |
| WinoGrande | — | 81.6% |
Math DeepSeek-R1-Distill-Qwen-32B leads
DeepSeek-R1-Distill-Qwen-32B: 34.5 (#194), GPT-3.5-turbo: 6.3 (#327)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 55.6% | 2.2% |
| FrontierMath (Tiers 1-3) | — | 0% |
| LiveBench Math | 59.4% | — |
| LMArena Math | — | 1142 |
| MATH Level 5 | — | 15.9% |
| GSM8K | — | 57.8% |
Knowledge DeepSeek-R1-Distill-Qwen-32B leads
DeepSeek-R1-Distill-Qwen-32B: 35.7 (#182), GPT-3.5-turbo: 10.0 (#303)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| GPQA Diamond | 64.1% | 28% |
| LMArena Expert | — | 1070 |
| ARC (AI2) Challenge | — | 87.4% |
| BoolQ | — | 87% |
| MMLU | — | 71.4% |
| OpenBookQA | — | 86% |
| TriviaQA | — | 85.8% |
Multilingual Not comparable
DeepSeek-R1-Distill-Qwen-32B: —, GPT-3.5-turbo: 31.5 (#258)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| LMArena Non-English | — | 1108 |
| LMArena Chinese | — | 1075 |
| LMArena French | — | 1118 |
| LMArena German | — | 1090 |
| LMArena Japanese | — | 1043 |
| LMArena Korean | — | 1019 |
| LMArena Russian | — | 1123 |
| LMArena Spanish | — | 1121 |
Instruction Following DeepSeek-R1-Distill-Qwen-32B leads
DeepSeek-R1-Distill-Qwen-32B: 61.6 (#243), GPT-3.5-turbo: 57.9 (#262)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| LiveBench Instruction Following | 55.7% | — |
| LMArena Instruction Following | — | 1119 |
Long Context Not comparable
DeepSeek-R1-Distill-Qwen-32B: —, GPT-3.5-turbo: 34.0 (#254)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| LMArena Longer Query | — | 1121 |
Writing & Preference DeepSeek-R1-Distill-Qwen-32B leads
DeepSeek-R1-Distill-Qwen-32B: 49.6 (#188), GPT-3.5-turbo: 25.3 (#305)
| Benchmark | DeepSeek-R1-Distill-Qwen-32B | GPT-3.5-turbo |
|---|---|---|
| LMArena Text | — | 1125 |
| LMArena Creative Writing | — | 1092 |
| EQ-Bench Creative Writing | — | 451 |
| LMArena Multi-Turn | — | 1117 |
| LiveBench Language | 26.8% | — |
Frequently asked questions
Is DeepSeek-R1-Distill-Qwen-32B better than GPT-3.5-turbo?
DeepSeek-R1-Distill-Qwen-32B is the stronger model overall, scoring 35.5 to 23.2 on the Noometry Index.
Is DeepSeek-R1-Distill-Qwen-32B or GPT-3.5-turbo better for coding?
DeepSeek-R1-Distill-Qwen-32B scores higher on coding benchmarks: 36.1 versus 23.9 in the Noometry coding category.
How many benchmarks do DeepSeek-R1-Distill-Qwen-32B and GPT-3.5-turbo share?
6 benchmarks have published results for both models. DeepSeek-R1-Distill-Qwen-32B has 14 scored results on Noometry and GPT-3.5-turbo has 44.