Model comparison
Claude 3.7 Sonnet vs DeepSeek-R1
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.5 on the Noometry Index.
Last verified . 44 shared benchmarks.
DeepSeekSummary
- They share 44 benchmarks with published results for both. Claude 3.7 Sonnet scores higher in 3 categories and DeepSeek-R1 in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in multilingual, where DeepSeek-R1 leads 52.4 to 44.1.
- The biggest single-benchmark swing is LiveBench Language: 59.9% for Claude 3.7 Sonnet and 48.5% for DeepSeek-R1.
Side by side
| Claude 3.7 Sonnet | DeepSeek-R1 | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 39.5 | 42.3 |
| Released | 2025-02-24 | 2025-01-20 |
| Weights | Proprietary | Proprietary |
| Context window | — | 164K |
| Max output | — | 64K |
| Input $ / M tokens | — | $0.50 |
| Output $ / M tokens | — | $2.15 |
| Results tracked | 58 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
Claude 3.7 Sonnet: 40.6 (#136), DeepSeek-R1: 46.3 (#68)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| Aider Polyglot | 64.9% | 71.4% |
| LiveBench Coding | 74.5% | 66.7% |
| LMArena Coding | 1361 | 1427 |
| SWE-bench Verified | 61% | — |
| SWE-bench Verified (bash only) | 52.8% | — |
| SciCode | — | 35.7% |
| GSO | 3.8% | — |
| WeirdML | — | 41.6% |
| CadEval | 54% | — |
| ALE-Bench | — | 804.12 |
| AlgoTune | — | 1.7 |
Agentic & Tool Use Claude 3.7 Sonnet leads
Claude 3.7 Sonnet: 34.1 (#50), DeepSeek-R1: 30.7 (#75)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| DeepResearch Bench | 43.6% | 35.1% |
| METR Time Horizons | 60% | 53.8% |
| TheAgentCompany | 30.9% | — |
| Cybench | 20% | — |
| OSWorld | 35.8% | — |
| BALROG | — | 34.9% |
Reasoning Too close to call
Claude 3.7 Sonnet: 18.6 (#277), DeepSeek-R1: 18.6 (#278)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| ARC-AGI-2 | 0.9% | 1.3% |
| SimpleBench | 46.4% | 40.8% |
| ARC-AGI-1 | 28.6% | 21.2% |
| LiveBench Reasoning | 87.8% | 83.2% |
| LMArena Hard Prompts | 1333 | 1416 |
| LiveBench Data Analysis | 74% | 69.8% |
| Epoch Capabilities Index | 141.16 | 141.29 |
| ForecastBench | 61.8 | 60 |
| LiveBench | 76.1% | 71.6% |
| Kagi LLM Benchmark | — | 69.4% |
| CritPt | — | 1.1% |
| EnigmaEval | 4.2% | — |
Math DeepSeek-R1 leads
Claude 3.7 Sonnet: 37.5 (#153), DeepSeek-R1: 43.8 (#79)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 57.8% | 66.4% |
| Omni-MATH | 33% | 42.4% |
| LiveBench Math | 79% | 80.7% |
| LMArena Math | 1337 | 1400 |
| MATH Level 5 | 91.2% | 96.6% |
| FrontierMath (Feb 2025 set) | 4.1% | — |
Knowledge DeepSeek-R1 leads
Claude 3.7 Sonnet: 39.8 (#130), DeepSeek-R1: 44.5 (#87)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| GPQA Diamond | 79.7% | 76.3% |
| MMLU-Pro | 78.4% | 79.3% |
| Confabulations | 14.7% | 12.7% |
| GPQA (HELM) | 60.8% | 66.6% |
| LMArena Expert | 1321 | 1394 |
| Humanity's Last Exam | 8% | — |
| Vectara Hallucination Rate | — | 11.3% |
Multimodal Not comparable
Claude 3.7 Sonnet: 33.7 (#95), DeepSeek-R1: —
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| LMArena Vision | 1169 | — |
| GeoBench | 68% | — |
| VPCT | 39% | — |
| SpatialViz-Bench | 33.9% | — |
Multilingual DeepSeek-R1 leads
Claude 3.7 Sonnet: 44.1 (#179), DeepSeek-R1: 52.4 (#85)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1296 | 1412 |
| LMArena Chinese | 1299 | 1442 |
| LMArena French | 1303 | 1417 |
| LMArena German | 1301 | 1404 |
| LMArena Japanese | 1267 | 1391 |
| LMArena Korean | 1249 | 1360 |
| LMArena Russian | 1311 | 1423 |
| LMArena Spanish | 1298 | 1411 |
Instruction Following Too close to call
Claude 3.7 Sonnet: 72.9 (#125), DeepSeek-R1: 72.0 (#143)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| LiveBench Instruction Following | 81.3% | 80.5% |
| IFEval | 83.4% | 78.4% |
| LMArena Instruction Following | 1352 | 1382 |
Long Context Claude 3.7 Sonnet leads
Claude 3.7 Sonnet: 50.3 (#10), DeepSeek-R1: 45.4 (#36)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| Fiction.LiveBench | 83.3% | 75% |
| LMArena Longer Query | 1373 | 1391 |
Writing & Preference DeepSeek-R1 leads
Claude 3.7 Sonnet: 54.4 (#150), DeepSeek-R1: 61.4 (#88)
| Benchmark | Claude 3.7 Sonnet | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1314 | 1428 |
| LMArena Creative Writing | 1332 | 1405 |
| Short-Story Creative Writing | 81.1% | 83% |
| EQ-Bench Creative Writing | 1412 | 1500 |
| WildBench | 81.4% | 82.8% |
| LMArena Multi-Turn | 1339 | 1405 |
| LiveBench Language | 59.9% | 48.5% |
Frequently asked questions
Is Claude 3.7 Sonnet better than DeepSeek-R1?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.5 on the Noometry Index.
Is Claude 3.7 Sonnet or DeepSeek-R1 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 40.6 in the Noometry coding category.
How many benchmarks do Claude 3.7 Sonnet and DeepSeek-R1 share?
44 benchmarks have published results for both models. Claude 3.7 Sonnet has 58 scored results on Noometry and DeepSeek-R1 has 52.