Model comparison
DeepSeek-R1 vs GPT-4o
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 28.6 on the Noometry Index.
Last verified . 46 shared benchmarks.
DeepSeekSummary
- They share 46 benchmarks with published results for both. DeepSeek-R1 scores higher in 9 categories and GPT-4o in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-R1 leads 43.8 to 10.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 66.4% for DeepSeek-R1 and 6.4% for GPT-4o.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- DeepSeek-R1 accepts more context: 164K tokens versus 128K.
Side by side
| DeepSeek-R1 | GPT-4o | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.3 | 28.6 |
| Released | 2025-01-20 | 2024-05-13 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | 128K |
| Max output | 64K | 16K |
| Input $ / M tokens | $0.50 | $2.50 |
| Output $ / M tokens | $2.15 | $10 |
| Results tracked | 52 | 72 |
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-4o: 24.8 (#328)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| Aider Polyglot | 71.4% | 45.3% |
| WeirdML | 41.6% | 25.1% |
| LiveBench Coding | 66.7% | 51.4% |
| LMArena Coding | 1427 | 1297 |
| SWE-bench Verified | — | 31% |
| SWE-bench Verified (bash only) | — | 21.6% |
| SciCode | 35.7% | — |
| GSO | — | 0% |
| BigCodeBench Instruct | — | 51.1% |
| BigCodeBench Complete | — | 61.1% |
| CadEval | — | 26% |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
| HumanEval+ | — | 87.2% |
| MBPP+ | — | 72.2% |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), GPT-4o: 21.0 (#141)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| BALROG | 34.9% | 32.3% |
| METR Time Horizons | 53.8% | 40.8% |
| GDPval | — | 9.9% |
| TheAgentCompany | — | 8.6% |
| Cybench | — | 12.5% |
| DeepResearch Bench | 35.1% | — |
| LMArena Search | — | 1006 |
Reasoning DeepSeek-R1 leads
DeepSeek-R1: 18.6 (#278), GPT-4o: 9.4 (#343)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| ARC-AGI-2 | 1.3% | 0% |
| SimpleBench | 40.8% | 17.8% |
| ARC-AGI-1 | 21.2% | 4.5% |
| CritPt | 1.1% | 0% |
| LiveBench Reasoning | 83.2% | 55.8% |
| LMArena Hard Prompts | 1416 | 1281 |
| LiveBench Data Analysis | 69.8% | 60.9% |
| Epoch Capabilities Index | 141.29 | 128.97 |
| ForecastBench | 60 | 57.7 |
| LiveBench | 71.6% | 55.3% |
| Kagi LLM Benchmark | 69.4% | — |
| Chess Puzzles | — | 13% |
| EnigmaEval | — | 0.8% |
| DTBench | — | 64.5% |
| LMCA | — | 16.6% |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), GPT-4o: 10.6 (#312)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 6.4% |
| Omni-MATH | 42.4% | 29.3% |
| LiveBench Math | 80.7% | 49.5% |
| LMArena Math | 1400 | 1285 |
| MATH Level 5 | 96.6% | 53.3% |
| FrontierMath (Tiers 1-3) | — | 0.4% |
| FrontierMath (Feb 2025 set) | — | 0.3% |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), GPT-4o: 28.8 (#242)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| GPQA Diamond | 76.3% | 49.2% |
| MMLU-Pro | 79.3% | 71.3% |
| Confabulations | 12.7% | 15.3% |
| Vectara Hallucination Rate | 11.3% | 9.6% |
| GPQA (HELM) | 66.6% | 52% |
| LMArena Expert | 1394 | 1250 |
| Humanity's Last Exam | — | 2.7% |
| SimpleQA Verified | — | 26% |
| MMLU | — | 88.1% |
Multimodal Not comparable
DeepSeek-R1: —, GPT-4o: 34.5 (#91)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| LMArena Vision | — | 1137 |
| Video-MME | — | 71.9% |
| GeoBench | — | 71% |
| VPCT | — | 40% |
| ScienceQA | — | 88.5% |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), GPT-4o: 43.2 (#186)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| LMArena Non-English | 1412 | 1283 |
| LMArena Chinese | 1442 | 1277 |
| LMArena French | 1417 | 1304 |
| LMArena German | 1404 | 1282 |
| LMArena Japanese | 1391 | 1257 |
| LMArena Korean | 1360 | 1234 |
| LMArena Russian | 1423 | 1286 |
| LMArena Spanish | 1411 | 1292 |
Instruction Following DeepSeek-R1 leads
DeepSeek-R1: 72.0 (#143), GPT-4o: 66.6 (#207)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| LiveBench Instruction Following | 80.5% | 68.6% |
| IFEval | 78.4% | 81.7% |
| LMArena Instruction Following | 1382 | 1278 |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), GPT-4o: 39.4 (#179)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| Fiction.LiveBench | 75% | 66.7% |
| LMArena Longer Query | 1391 | 1289 |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), GPT-4o: 52.6 (#166)
| Benchmark | DeepSeek-R1 | GPT-4o |
|---|---|---|
| LMArena Text | 1428 | 1300 |
| LMArena Creative Writing | 1405 | 1292 |
| Short-Story Creative Writing | 83% | 81.8% |
| WildBench | 82.8% | 82.8% |
| LMArena Multi-Turn | 1405 | 1302 |
| LiveBench Language | 48.5% | 47.6% |
| EQ-Bench Creative Writing | 1500 | — |
Frequently asked questions
Is DeepSeek-R1 better than GPT-4o?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 28.6 on the Noometry Index.
Which is cheaper, DeepSeek-R1 or GPT-4o?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; GPT-4o lists at $2.50 and $10.
Is DeepSeek-R1 or GPT-4o better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 24.8 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 128K.
How many benchmarks do DeepSeek-R1 and GPT-4o share?
46 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and GPT-4o has 72.