Model comparison
Claude Opus 4 vs DeepSeek-R1
Claude Opus 4 and DeepSeek-R1 score almost the same on the Noometry Index (43.1 vs 42.3), so choose on price, context window or the category you care about most.
Last verified . 42 shared benchmarks.
DeepSeekSummary
- They share 42 benchmarks with published results for both. Claude Opus 4 scores higher in 4 categories and DeepSeek-R1 in 5 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Opus 4 leads 27.3 to 18.6.
- The biggest single-benchmark swing is Omni-MATH: 61.6% for Claude Opus 4 and 42.4% for DeepSeek-R1.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $15 / $75 for Claude Opus 4.
- Claude Opus 4 accepts more context: 200K tokens versus 164K.
Side by side
| Claude Opus 4 | DeepSeek-R1 | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 43.1 | 42.3 |
| Released | 2025-05-22 | 2025-01-20 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 164K |
| Max output | 32K | 64K |
| Input $ / M tokens | $15 | $0.50 |
| Output $ / M tokens | $75 | $2.15 |
| Results tracked | 56 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Claude Opus 4: 47.2 (#62), DeepSeek-R1: 46.3 (#68)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| Aider Polyglot | 72% | 71.4% |
| WeirdML | 43.7% | 41.6% |
| LMArena Coding | 1442 | 1427 |
| AlgoTune | 1.33 | 1.7 |
| SWE-bench Verified | 70.7% | — |
| SWE-bench Verified (bash only) | 67.6% | — |
| SciCode | — | 35.7% |
| GSO | 6.9% | — |
| LiveBench Coding | — | 66.7% |
| ALE-Bench | — | 804.12 |
Agentic & Tool Use Claude Opus 4 leads
Claude Opus 4: 34.8 (#42), DeepSeek-R1: 30.7 (#75)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| DeepResearch Bench | 46.8% | 35.1% |
| METR Time Horizons | 63.9% | 53.8% |
| Cybench | 38% | — |
| BALROG | — | 34.9% |
| LMArena Search | 1127 | — |
Reasoning Claude Opus 4 leads
Claude Opus 4: 27.3 (#121), DeepSeek-R1: 18.6 (#278)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| ARC-AGI-2 | 8.6% | 1.3% |
| SimpleBench | 58.8% | 40.8% |
| Kagi LLM Benchmark | 74.3% | 69.4% |
| ARC-AGI-1 | 35.7% | 21.2% |
| CritPt | 0.3% | 1.1% |
| LMArena Hard Prompts | 1399 | 1416 |
| Epoch Capabilities Index | 142.67 | 141.29 |
| ForecastBench | 61.1 | 60 |
| EnigmaEval | 5.6% | — |
| LiveBench Reasoning | — | 83.2% |
| DTBench | 81.6% | — |
| LiveBench Data Analysis | — | 69.8% |
| LMCA | 37.4% | — |
| LiveBench | — | 71.6% |
Math DeepSeek-R1 leads
Claude Opus 4: 42.0 (#86), DeepSeek-R1: 43.8 (#79)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 64.4% | 66.4% |
| Omni-MATH | 61.6% | 42.4% |
| LMArena Math | 1390 | 1400 |
| MATH Level 5 | 85% | 96.6% |
| LiveBench Math | — | 80.7% |
| FrontierMath (Feb 2025 set) | 4.5% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |
Knowledge Too close to call
Claude Opus 4: 44.0 (#88), DeepSeek-R1: 44.5 (#87)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| GPQA Diamond | 76.3% | 76.3% |
| MMLU-Pro | 87.5% | 79.3% |
| Confabulations | 15.9% | 12.7% |
| Vectara Hallucination Rate | 12% | 11.3% |
| GPQA (HELM) | 70.8% | 66.6% |
| LMArena Expert | 1386 | 1394 |
| Humanity's Last Exam | 10.7% | — |
Multimodal Not comparable
Claude Opus 4: 31.5 (#106), DeepSeek-R1: —
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| LMArena Vision | 1192 | — |
| GeoBench | 49% | — |
| VPCT | 38% | — |
Multilingual DeepSeek-R1 leads
Claude Opus 4: 48.8 (#138), DeepSeek-R1: 52.4 (#85)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1362 | 1412 |
| LMArena Chinese | 1386 | 1442 |
| LMArena French | 1372 | 1417 |
| LMArena German | 1391 | 1404 |
| LMArena Japanese | 1331 | 1391 |
| LMArena Korean | 1321 | 1360 |
| LMArena Russian | 1392 | 1423 |
| LMArena Spanish | 1389 | 1411 |
Instruction Following Claude Opus 4 leads
Claude Opus 4: 77.1 (#28), DeepSeek-R1: 72.0 (#143)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| IFEval | 91.8% | 78.4% |
| LMArena Instruction Following | 1406 | 1382 |
| LiveBench Instruction Following | — | 80.5% |
Long Context DeepSeek-R1 leads
Claude Opus 4: 39.6 (#172), DeepSeek-R1: 45.4 (#36)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| Fiction.LiveBench | 61.1% | 75% |
| LMArena Longer Query | 1422 | 1391 |
Writing & Preference Too close to call
Claude Opus 4: 61.2 (#89), DeepSeek-R1: 61.4 (#88)
| Benchmark | Claude Opus 4 | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1377 | 1428 |
| LMArena Creative Writing | 1387 | 1405 |
| Short-Story Creative Writing | 83.6% | 83% |
| EQ-Bench Creative Writing | 1580 | 1500 |
| WildBench | 85.2% | 82.8% |
| LMArena Multi-Turn | 1396 | 1405 |
| LiveBench Language | — | 48.5% |
Frequently asked questions
Is Claude Opus 4 better than DeepSeek-R1?
Claude Opus 4 and DeepSeek-R1 score almost the same on the Noometry Index (43.1 vs 42.3), so choose on price, context window or the category you care about most.
Which is cheaper, Claude Opus 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 Opus 4 lists at $15 and $75.
Is Claude Opus 4 or DeepSeek-R1 better for coding?
They score almost the same on coding (47.2 vs 46.3); test both on your own repository before choosing.
Which has the bigger context window?
Claude Opus 4 does, with 200K tokens against 164K.
How many benchmarks do Claude Opus 4 and DeepSeek-R1 share?
42 benchmarks have published results for both models. Claude Opus 4 has 56 scored results on Noometry and DeepSeek-R1 has 52.