Model comparison
DeepSeek-R1 vs GPT-4.5
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 37.2 on the Noometry Index.
Last verified . 37 shared benchmarks.
DeepSeekSummary
- They share 37 benchmarks with published results for both. DeepSeek-R1 scores higher in 7 categories and GPT-4.5 in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-R1 leads 44.5 to 32.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 66.4% for DeepSeek-R1 and 37.8% for GPT-4.5.
Side by side
| DeepSeek-R1 | GPT-4.5 | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.3 | 37.2 |
| Released | 2025-01-20 | 2025-02-27 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | — |
| Max output | 64K | — |
| Input $ / M tokens | $0.50 | — |
| Output $ / M tokens | $2.15 | — |
| Results tracked | 52 | 42 |
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-4.5: 42.2 (#109)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| Aider Polyglot | 71.4% | 44.9% |
| WeirdML | 41.6% | 39.4% |
| LiveBench Coding | 66.7% | 75.2% |
| LMArena Coding | 1427 | 1396 |
| SciCode | 35.7% | — |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), GPT-4.5: 27.9 (#97)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| Cybench | — | 17.5% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning DeepSeek-R1 leads
DeepSeek-R1: 18.6 (#278), GPT-4.5: 13.9 (#330)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| ARC-AGI-2 | 1.3% | 0.8% |
| SimpleBench | 40.8% | 34.5% |
| ARC-AGI-1 | 21.2% | 10.3% |
| LiveBench Reasoning | 83.2% | 71.1% |
| LMArena Hard Prompts | 1416 | 1403 |
| LiveBench Data Analysis | 69.8% | 64.3% |
| Epoch Capabilities Index | 141.29 | 136.74 |
| ForecastBench | 60 | 61.7 |
| LiveBench | 71.6% | 69% |
| Kagi LLM Benchmark | 69.4% | — |
| CritPt | 1.1% | — |
| EnigmaEval | — | 3.2% |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), GPT-4.5: 32.6 (#211)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 37.8% |
| LiveBench Math | 80.7% | 69.3% |
| LMArena Math | 1400 | 1412 |
| MATH Level 5 | 96.6% | 78.6% |
| Omni-MATH | 42.4% | — |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), GPT-4.5: 32.5 (#211)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| GPQA Diamond | 76.3% | 68.7% |
| Confabulations | 12.7% | 13.6% |
| LMArena Expert | 1394 | 1394 |
| Humanity's Last Exam | — | 5.4% |
| MMLU-Pro | 79.3% | — |
| Vectara Hallucination Rate | 11.3% | — |
| GPQA (HELM) | 66.6% | — |
Multimodal Not comparable
DeepSeek-R1: —, GPT-4.5: 37.6 (#71)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| LMArena Vision | — | 1195 |
| VPCT | — | 45% |
Multilingual Too close to call
DeepSeek-R1: 52.4 (#85), GPT-4.5: 52.5 (#83)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| LMArena Non-English | 1412 | 1413 |
| LMArena Chinese | 1442 | 1421 |
| LMArena French | 1417 | 1418 |
| LMArena German | 1404 | 1457 |
| LMArena Japanese | 1391 | 1416 |
| LMArena Korean | 1360 | 1392 |
| LMArena Russian | 1423 | 1419 |
| LMArena Spanish | 1411 | — |
Instruction Following Too close to call
DeepSeek-R1: 72.0 (#143), GPT-4.5: 72.6 (#134)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| LiveBench Instruction Following | 80.5% | 72.3% |
| LMArena Instruction Following | 1382 | 1404 |
| IFEval | 78.4% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), GPT-4.5: 40.4 (#155)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| Fiction.LiveBench | 75% | 63.9% |
| LMArena Longer Query | 1391 | 1406 |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), GPT-4.5: 56.9 (#134)
| Benchmark | DeepSeek-R1 | GPT-4.5 |
|---|---|---|
| LMArena Text | 1428 | 1417 |
| LMArena Creative Writing | 1405 | 1394 |
| Short-Story Creative Writing | 83% | 75.6% |
| EQ-Bench Creative Writing | 1500 | 1258 |
| LMArena Multi-Turn | 1405 | 1444 |
| LiveBench Language | 48.5% | 61.5% |
| WildBench | 82.8% | — |
Frequently asked questions
Is DeepSeek-R1 better than GPT-4.5?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 37.2 on the Noometry Index.
Is DeepSeek-R1 or GPT-4.5 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 42.2 in the Noometry coding category.
How many benchmarks do DeepSeek-R1 and GPT-4.5 share?
37 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and GPT-4.5 has 42.