Model comparison
Claude 3.5 Sonnet vs DeepSeek-R1
DeepSeek-R1 is the stronger model overall, scoring 42.0 to 34.4 on the Noometry Index.
Last verified . 42 shared benchmarks.
DeepSeekSummary
- They share 42 benchmarks with published results for both. Claude 3.5 Sonnet scores higher in 2 categories and DeepSeek-R1 in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-R1 leads 42.0 to 18.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 8.5% for Claude 3.5 Sonnet and 66.4% for DeepSeek-R1.
Side by side
| Claude 3.5 Sonnet | DeepSeek-R1 | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 34.4 | 42.0 |
| Released | 2024-06-20 | 2025-01-20 |
| Weights | Proprietary | Proprietary |
| Context window | — | 164K |
| Max output | — | 64K |
| Input $ / M tokens | — | $0.50 |
| Output $ / M tokens | — | $2.15 |
| Results tracked | 60 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
Claude 3.5 Sonnet: 39.5 (#145), DeepSeek-R1: 45.9 (#63)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| Aider Polyglot | 51.6% | 71.4% |
| WeirdML | 40% | 41.6% |
| LiveBench Coding | 67.1% | 66.7% |
| LMArena Coding | 1343 | 1427 |
| SciCode | — | 35.7% |
| GSO | 4.6% | — |
| BigCodeBench Instruct | 46.8% | — |
| BigCodeBench Complete | 58.6% | — |
| CadEval | 48% | — |
| ALE-Bench | — | 804.12 |
| AlgoTune | — | 1.7 |
| HumanEval+ | 81.7% | — |
| MBPP+ | 74.3% | — |
Agentic & Tool Use Claude 3.5 Sonnet leads
Claude 3.5 Sonnet: 32.1 (#61), DeepSeek-R1: 30.4 (#68)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| BALROG | 32.6% | 34.9% |
| METR Time Horizons | 45.2% | 53.8% |
| TheAgentCompany | 24% | — |
| Cybench | 17.5% | — |
| DeepResearch Bench | — | 35.1% |
Reasoning Claude 3.5 Sonnet leads
Claude 3.5 Sonnet: 22.5 (#176), DeepSeek-R1: 18.7 (#253)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| SimpleBench | 41.4% | 40.8% |
| LiveBench Reasoning | 56.7% | 83.2% |
| LMArena Hard Prompts | 1305 | 1415 |
| LiveBench Data Analysis | 55% | 69.8% |
| Epoch Capabilities Index | 133.55 | 141.3 |
| ForecastBench | 60.7 | 60 |
| LiveBench | 59% | 71.6% |
| ARC-AGI-2 | — | 1.3% |
| Kagi LLM Benchmark | — | 69.4% |
| ARC-AGI-1 | — | 21.2% |
| CritPt | — | 1.1% |
| EnigmaEval | 0.9% | — |
| DTBench | 67.8% | — |
Math DeepSeek-R1 leads
Claude 3.5 Sonnet: 18.0 (#282), DeepSeek-R1: 42.0 (#70)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 8.5% | 66.4% |
| Omni-MATH | 27.6% | 42.4% |
| LiveBench Math | 52.3% | 80.7% |
| LMArena Math | 1306 | 1401 |
| MATH Level 5 | 56.9% | 96.6% |
| FrontierMath (Feb 2025 set) | 2.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge DeepSeek-R1 leads
Claude 3.5 Sonnet: 28.5 (#236), DeepSeek-R1: 44.3 (#81)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| GPQA Diamond | 55.3% | 76.3% |
| MMLU-Pro | 77.7% | 79.3% |
| Confabulations | 19.9% | 12.7% |
| GPQA (HELM) | 56.5% | 66.6% |
| LMArena Expert | 1264 | 1394 |
| Humanity's Last Exam | 4.1% | — |
| Vectara Hallucination Rate | — | 11.3% |
| MMLU | 87.3% | — |
Multimodal Not comparable
Claude 3.5 Sonnet: 26.0 (#115), DeepSeek-R1: —
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| LMArena Vision | 1125 | — |
| Video-MME | 60% | — |
| GeoBench | 62% | — |
| VPCT | 33% | — |
Multilingual DeepSeek-R1 leads
Claude 3.5 Sonnet: 43.4 (#180), DeepSeek-R1: 52.6 (#80)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1283 | 1412 |
| LMArena Chinese | 1271 | 1443 |
| LMArena French | 1304 | 1415 |
| LMArena German | 1297 | 1405 |
| LMArena Japanese | 1234 | 1394 |
| LMArena Korean | 1200 | 1359 |
| LMArena Russian | 1305 | 1423 |
| LMArena Spanish | 1289 | 1410 |
Instruction Following DeepSeek-R1 leads
Claude 3.5 Sonnet: 69.0 (#177), DeepSeek-R1: 72.2 (#138)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| LiveBench Instruction Following | 69.3% | 80.5% |
| IFEval | 85.5% | 78.4% |
| LMArena Instruction Following | 1297 | 1382 |
Long Context DeepSeek-R1 leads
Claude 3.5 Sonnet: 39.9 (#162), DeepSeek-R1: 45.4 (#36)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| LMArena Longer Query | 1311 | 1391 |
| Fiction.LiveBench | — | 75% |
Writing & Preference DeepSeek-R1 leads
Claude 3.5 Sonnet: 53.0 (#159), DeepSeek-R1: 61.5 (#84)
| Benchmark | Claude 3.5 Sonnet | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1298 | 1428 |
| LMArena Creative Writing | 1292 | 1406 |
| Short-Story Creative Writing | 80.3% | 83% |
| EQ-Bench Creative Writing | 1451 | 1500 |
| WildBench | 79.2% | 82.8% |
| LMArena Multi-Turn | 1326 | 1404 |
| LiveBench Language | 53.8% | 48.5% |
Frequently asked questions
Is Claude 3.5 Sonnet better than DeepSeek-R1?
DeepSeek-R1 is the stronger model overall, scoring 42.0 to 34.4 on the Noometry Index.
Is Claude 3.5 Sonnet or DeepSeek-R1 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 45.9 versus 39.5 in the Noometry coding category.
How many benchmarks do Claude 3.5 Sonnet and DeepSeek-R1 share?
42 benchmarks have published results for both models. Claude 3.5 Sonnet has 60 scored results on Noometry and DeepSeek-R1 has 52.