Model comparison
DeepSeek-R1 vs GPT-3.5-turbo
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 23.2 on the Noometry Index.
Last verified . 25 shared benchmarks.
DeepSeekSummary
- They share 25 benchmarks with published results for both. DeepSeek-R1 scores higher in 8 categories and GPT-3.5-turbo in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-R1 leads 43.8 to 6.3.
- The biggest single-benchmark swing is MATH Level 5: 96.6% for DeepSeek-R1 and 15.9% for GPT-3.5-turbo.
- GPT-3.5-turbo is cheaper at $0.50 / $1.50 per million input/output tokens, against $0.50 / $2.15 for DeepSeek-R1.
- DeepSeek-R1 accepts more context: 164K tokens versus 16K.
Side by side
| DeepSeek-R1 | GPT-3.5-turbo | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.3 | 23.2 |
| Released | 2025-01-20 | 2023-03-01 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | 16K |
| Max output | 64K | 4K |
| Input $ / M tokens | $0.50 | $0.50 |
| Output $ / M tokens | $2.15 | $1.50 |
| Results tracked | 52 | 44 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), GPT-3.5-turbo: 23.9 (#331)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| WeirdML | 41.6% | 3.5% |
| LMArena Coding | 1427 | 1136 |
| Aider Polyglot | 71.4% | — |
| SciCode | 35.7% | — |
| BigCodeBench Instruct | — | 39.1% |
| LiveBench Coding | 66.7% | — |
| BigCodeBench Complete | — | 50.6% |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
| HumanEval+ | — | 70.7% |
| MBPP+ | — | 69.7% |
Agentic & Tool Use Not comparable
DeepSeek-R1: 30.7 (#75), GPT-3.5-turbo: —
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| METR Time Horizons | 53.8% | 21.5% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
Reasoning DeepSeek-R1 leads
DeepSeek-R1: 18.6 (#278), GPT-3.5-turbo: 13.8 (#332)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| LMArena Hard Prompts | 1416 | 1108 |
| Epoch Capabilities Index | 141.29 | 118.55 |
| ForecastBench | 60 | 50.4 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| Kagi LLM Benchmark | 69.4% | — |
| ARC-AGI-1 | 21.2% | — |
| CritPt | 1.1% | — |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | 83.2% | — |
| Mystery Game Puzzles | — | 3% |
| DTBench | — | 48.5% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 9.7% |
| Adversarial NLI | — | 58.1% |
| BIG-Bench Hard | — | 61.6% |
| CommonsenseQA 2.0 | — | 57% |
| LiveBench | 71.6% | — |
| WinoGrande | — | 81.6% |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), GPT-3.5-turbo: 6.3 (#327)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 2.2% |
| LMArena Math | 1400 | 1142 |
| MATH Level 5 | 96.6% | 15.9% |
| FrontierMath (Tiers 1-3) | — | 0% |
| Omni-MATH | 42.4% | — |
| LiveBench Math | 80.7% | — |
| GSM8K | — | 57.8% |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), GPT-3.5-turbo: 10.0 (#303)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| GPQA Diamond | 76.3% | 28% |
| LMArena Expert | 1394 | 1070 |
| MMLU-Pro | 79.3% | — |
| Confabulations | 12.7% | — |
| Vectara Hallucination Rate | 11.3% | — |
| GPQA (HELM) | 66.6% | — |
| ARC (AI2) Challenge | — | 87.4% |
| BoolQ | — | 87% |
| MMLU | — | 71.4% |
| OpenBookQA | — | 86% |
| TriviaQA | — | 85.8% |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), GPT-3.5-turbo: 31.5 (#258)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| LMArena Non-English | 1412 | 1108 |
| LMArena Chinese | 1442 | 1075 |
| LMArena French | 1417 | 1118 |
| LMArena German | 1404 | 1090 |
| LMArena Japanese | 1391 | 1043 |
| LMArena Korean | 1360 | 1019 |
| LMArena Russian | 1423 | 1123 |
| LMArena Spanish | 1411 | 1121 |
Instruction Following DeepSeek-R1 leads
DeepSeek-R1: 72.0 (#143), GPT-3.5-turbo: 57.9 (#262)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| LMArena Instruction Following | 1382 | 1119 |
| LiveBench Instruction Following | 80.5% | — |
| IFEval | 78.4% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), GPT-3.5-turbo: 34.0 (#254)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| LMArena Longer Query | 1391 | 1121 |
| Fiction.LiveBench | 75% | — |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), GPT-3.5-turbo: 25.3 (#305)
| Benchmark | DeepSeek-R1 | GPT-3.5-turbo |
|---|---|---|
| LMArena Text | 1428 | 1125 |
| LMArena Creative Writing | 1405 | 1092 |
| EQ-Bench Creative Writing | 1500 | 451 |
| LMArena Multi-Turn | 1405 | 1117 |
| Short-Story Creative Writing | 83% | — |
| WildBench | 82.8% | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than GPT-3.5-turbo?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 23.2 on the Noometry Index.
Which is cheaper, DeepSeek-R1 or GPT-3.5-turbo?
GPT-3.5-turbo is cheaper. It lists at $0.50 per million input tokens and $1.50 per million output tokens; DeepSeek-R1 lists at $0.50 and $2.15.
Is DeepSeek-R1 or GPT-3.5-turbo better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 23.9 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 16K.
How many benchmarks do DeepSeek-R1 and GPT-3.5-turbo share?
25 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and GPT-3.5-turbo has 44.