Model comparison
DeepSeek-R1 vs Grok 3
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.9 on the Noometry Index.
Last verified . 38 shared benchmarks.
DeepSeekSummary
- They share 38 benchmarks with published results for both. DeepSeek-R1 scores higher in 7 categories and Grok 3 in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in long context, where DeepSeek-R1 leads 45.4 to 38.7.
- The biggest single-benchmark swing is Aider Polyglot: 71.4% for DeepSeek-R1 and 53.3% for Grok 3.
Side by side
| DeepSeek-R1 | Grok 3 | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 42.3 | 39.9 |
| Released | 2025-01-20 | 2025-04-09 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | — |
| Max output | 64K | — |
| Input $ / M tokens | $0.50 | — |
| Output $ / M tokens | $2.15 | — |
| Results tracked | 52 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), Grok 3: 41.9 (#115)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| Aider Polyglot | 71.4% | 53.3% |
| WeirdML | 41.6% | 37.2% |
| LMArena Coding | 1427 | 1432 |
| SciCode | 35.7% | — |
| LiveBench Coding | 66.7% | — |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use Too close to call
DeepSeek-R1: 30.7 (#75), Grok 3: 30.5 (#76)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| BALROG | 34.9% | 29.5% |
| DeepResearch Bench | 35.1% | — |
| METR Time Horizons | 53.8% | — |
Reasoning DeepSeek-R1 leads
DeepSeek-R1: 18.6 (#278), Grok 3: 13.7 (#333)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| ARC-AGI-2 | 1.3% | 0% |
| SimpleBench | 40.8% | 36.1% |
| Kagi LLM Benchmark | 69.4% | 61.3% |
| ARC-AGI-1 | 21.2% | 5.5% |
| LMArena Hard Prompts | 1416 | 1434 |
| Epoch Capabilities Index | 141.29 | 138.33 |
| CritPt | 1.1% | — |
| LiveBench Reasoning | 83.2% | — |
| LiveBench Data Analysis | 69.8% | — |
| ForecastBench | 60 | — |
| LiveBench | 71.6% | — |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), Grok 3: 38.0 (#145)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 55.6% |
| Omni-MATH | 42.4% | 46.4% |
| LMArena Math | 1400 | 1391 |
| MATH Level 5 | 96.6% | 88.7% |
| LiveBench Math | 80.7% | — |
| FrontierMath (Feb 2025 set) | — | 3.8% |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Grok 3 leads
DeepSeek-R1: 44.5 (#87), Grok 3: 46.2 (#82)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| GPQA Diamond | 76.3% | 75.8% |
| MMLU-Pro | 79.3% | 78.8% |
| Confabulations | 12.7% | 14.2% |
| Vectara Hallucination Rate | 11.3% | 5.8% |
| GPQA (HELM) | 66.6% | 65% |
| LMArena Expert | 1394 | 1421 |
Multilingual Too close to call
DeepSeek-R1: 52.4 (#85), Grok 3: 52.3 (#87)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| LMArena Non-English | 1412 | 1410 |
| LMArena Chinese | 1442 | 1448 |
| LMArena French | 1417 | 1460 |
| LMArena German | 1404 | 1431 |
| LMArena Japanese | 1391 | 1387 |
| LMArena Korean | 1360 | 1373 |
| LMArena Russian | 1423 | 1416 |
| LMArena Spanish | 1411 | 1417 |
Instruction Following Grok 3 leads
DeepSeek-R1: 72.0 (#143), Grok 3: 75.0 (#73)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| IFEval | 78.4% | 88.4% |
| LMArena Instruction Following | 1382 | 1409 |
| LiveBench Instruction Following | 80.5% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), Grok 3: 38.7 (#192)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| Fiction.LiveBench | 75% | 58.3% |
| LMArena Longer Query | 1391 | 1439 |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), Grok 3: 55.8 (#141)
| Benchmark | DeepSeek-R1 | Grok 3 |
|---|---|---|
| LMArena Text | 1428 | 1426 |
| LMArena Creative Writing | 1405 | 1414 |
| Short-Story Creative Writing | 83% | 76.4% |
| EQ-Bench Creative Writing | 1500 | 1186 |
| WildBench | 82.8% | 84.9% |
| LMArena Multi-Turn | 1405 | 1425 |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than Grok 3?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.9 on the Noometry Index.
Is DeepSeek-R1 or Grok 3 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 41.9 in the Noometry coding category.
How many benchmarks do DeepSeek-R1 and Grok 3 share?
38 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Grok 3 has 40.