Model comparison
Claude Sonnet 4 vs DeepSeek-R1
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 40.8 on the Noometry Index.
Last verified . 43 shared benchmarks.
DeepSeekSummary
- They share 43 benchmarks with published results for both. Claude Sonnet 4 scores higher in 2 categories and DeepSeek-R1 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in long context, where DeepSeek-R1 leads 45.4 to 33.7.
- The biggest single-benchmark swing is Fiction.LiveBench: 46.9% for Claude Sonnet 4 and 75% for DeepSeek-R1.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- Claude Sonnet 4 accepts more context: 200K tokens versus 164K.
Side by side
| Claude Sonnet 4 | DeepSeek-R1 | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 40.8 | 42.3 |
| Released | 2025-05-22 | 2025-01-20 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 164K |
| Max output | 64K | 64K |
| Input $ / M tokens | $3 | $0.50 |
| Output $ / M tokens | $15 | $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 Sonnet 4: 43.5 (#88), DeepSeek-R1: 46.3 (#68)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| Aider Polyglot | 61.3% | 71.4% |
| SciCode | 40% | 35.7% |
| WeirdML | 46.1% | 41.6% |
| LMArena Coding | 1414 | 1427 |
| ALE-Bench | 655.35 | 804.12 |
| SWE-bench Verified (bash only) | 64.9% | — |
| GSO | 4.9% | — |
| LiveBench Coding | — | 66.7% |
| AlgoTune | — | 1.7 |
Agentic & Tool Use Claude Sonnet 4 leads
Claude Sonnet 4: 38.5 (#31), DeepSeek-R1: 30.7 (#75)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| DeepResearch Bench | 46.6% | 35.1% |
| METR Time Horizons | 62% | 53.8% |
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| OSWorld | 43.9% | — |
| BALROG | — | 34.9% |
Reasoning Claude Sonnet 4 leads
Claude Sonnet 4: 22.9 (#187), DeepSeek-R1: 18.6 (#278)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| ARC-AGI-2 | 5.9% | 1.3% |
| SimpleBench | 45.5% | 40.8% |
| Kagi LLM Benchmark | 73% | 69.4% |
| ARC-AGI-1 | 40% | 21.2% |
| CritPt | 0.3% | 1.1% |
| LMArena Hard Prompts | 1372 | 1416 |
| Epoch Capabilities Index | 141.69 | 141.29 |
| ForecastBench | 60.2 | 60 |
| EnigmaEval | 3.1% | — |
| LiveBench Reasoning | — | 83.2% |
| DTBench | 77.1% | — |
| LiveBench Data Analysis | — | 69.8% |
| LMCA | 29% | — |
| LiveBench | — | 71.6% |
Math Too close to call
Claude Sonnet 4: 43.3 (#80), DeepSeek-R1: 43.8 (#79)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 66.4% |
| Omni-MATH | 60.2% | 42.4% |
| LMArena Math | 1375 | 1400 |
| MATH Level 5 | 84.4% | 96.6% |
| LiveBench Math | — | 80.7% |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge DeepSeek-R1 leads
Claude Sonnet 4: 41.8 (#108), DeepSeek-R1: 44.5 (#87)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| GPQA Diamond | 79.2% | 76.3% |
| MMLU-Pro | 84.3% | 79.3% |
| Confabulations | 13.2% | 12.7% |
| Vectara Hallucination Rate | 10.3% | 11.3% |
| GPQA (HELM) | 70.6% | 66.6% |
| LMArena Expert | 1372 | 1394 |
| Humanity's Last Exam | 7.8% | — |
Multimodal Not comparable
Claude Sonnet 4: 26.2 (#121), DeepSeek-R1: —
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| LMArena Vision | 1191 | — |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |
Multilingual DeepSeek-R1 leads
Claude Sonnet 4: 46.7 (#156), DeepSeek-R1: 52.4 (#85)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1333 | 1412 |
| LMArena Chinese | 1350 | 1442 |
| LMArena French | 1363 | 1417 |
| LMArena German | 1331 | 1404 |
| LMArena Japanese | 1302 | 1391 |
| LMArena Korean | 1291 | 1360 |
| LMArena Russian | 1355 | 1423 |
| LMArena Spanish | 1357 | 1411 |
Instruction Following Too close to call
Claude Sonnet 4: 71.7 (#145), DeepSeek-R1: 72.0 (#143)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| IFEval | 84% | 78.4% |
| LMArena Instruction Following | 1376 | 1382 |
| LiveBench Instruction Following | — | 80.5% |
Long Context DeepSeek-R1 leads
Claude Sonnet 4: 33.7 (#259), DeepSeek-R1: 45.4 (#36)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| Fiction.LiveBench | 46.9% | 75% |
| LMArena Longer Query | 1398 | 1391 |
Writing & Preference DeepSeek-R1 leads
Claude Sonnet 4: 57.1 (#132), DeepSeek-R1: 61.4 (#88)
| Benchmark | Claude Sonnet 4 | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1351 | 1428 |
| LMArena Creative Writing | 1345 | 1405 |
| Short-Story Creative Writing | 81.4% | 83% |
| EQ-Bench Creative Writing | 1483 | 1500 |
| WildBench | 83.8% | 82.8% |
| LMArena Multi-Turn | 1376 | 1405 |
| LiveBench Language | — | 48.5% |
Frequently asked questions
Is Claude Sonnet 4 better than DeepSeek-R1?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 40.8 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 or DeepSeek-R1?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Claude Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or DeepSeek-R1 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 43.5 in the Noometry coding category.
Which has the bigger context window?
Claude Sonnet 4 does, with 200K tokens against 164K.
How many benchmarks do Claude Sonnet 4 and DeepSeek-R1 share?
43 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and DeepSeek-R1 has 52.