Model comparison
DeepSeek-V3.1 vs Grok 3
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 39.9 on the Noometry Index.
Last verified . 24 shared benchmarks.
DeepSeekSummary
- They share 24 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 3 categories and Grok 3 in 5 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek-V3.1 leads 27.9 to 13.7.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 53.2% for DeepSeek-V3.1 and 61.3% for Grok 3.
- DeepSeek-V3.1 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.1 | Grok 3 | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 42.8 | 39.9 |
| Released | 2025-08-21 | 2025-04-09 |
| Weights | Open | Proprietary |
| Context window | 164K | — |
| Max output | 8K | — |
| Input $ / M tokens | $0.25 | — |
| Output $ / M tokens | $0.95 | — |
| Results tracked | 27 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 3 leads
DeepSeek-V3.1: 40.3 (#144), Grok 3: 41.9 (#115)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| WeirdML | 38.4% | 37.2% |
| LMArena Coding | 1417 | 1432 |
| Aider Polyglot | — | 53.3% |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, Grok 3: 30.5 (#76)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| BALROG | — | 29.5% |
Reasoning DeepSeek-V3.1 leads
DeepSeek-V3.1: 27.9 (#110), Grok 3: 13.7 (#333)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| SimpleBench | 40% | 36.1% |
| Kagi LLM Benchmark | 53.2% | 61.3% |
| LMArena Hard Prompts | 1417 | 1434 |
| Epoch Capabilities Index | 139.92 | 138.33 |
| ARC-AGI-2 | — | 0% |
| ARC-AGI-1 | — | 5.5% |
| DTBench | 82.7% | — |
| LMCA | 24.3% | — |
| ForecastBench | 58 | — |
Math Too close to call
DeepSeek-V3.1: 38.9 (#122), Grok 3: 38.0 (#145)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| LMArena Math | 1420 | 1391 |
| OTIS Mock AIME 2024-2025 | — | 55.6% |
| Omni-MATH | — | 46.4% |
| MATH Level 5 | — | 88.7% |
| FrontierMath (Feb 2025 set) | — | 3.8% |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Grok 3 leads
DeepSeek-V3.1: 43.7 (#90), Grok 3: 46.2 (#82)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| Vectara Hallucination Rate | 5.5% | 5.8% |
| LMArena Expert | 1405 | 1421 |
| GPQA Diamond | — | 75.8% |
| MMLU-Pro | — | 78.8% |
| Confabulations | — | 14.2% |
| GPQA (HELM) | — | 65% |
Multilingual Too close to call
DeepSeek-V3.1: 51.6 (#106), Grok 3: 52.3 (#87)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| LMArena Non-English | 1400 | 1410 |
| LMArena Chinese | 1469 | 1448 |
| LMArena French | 1447 | 1460 |
| LMArena German | 1411 | 1431 |
| LMArena Japanese | 1378 | 1387 |
| LMArena Korean | 1337 | 1373 |
| LMArena Russian | 1405 | 1416 |
| LMArena Spanish | 1431 | 1417 |
Instruction Following Grok 3 leads
DeepSeek-V3.1: 73.9 (#110), Grok 3: 75.0 (#73)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| LMArena Instruction Following | 1400 | 1409 |
| IFEval | — | 88.4% |
Long Context Grok 3 leads
DeepSeek-V3.1: 36.3 (#232), Grok 3: 38.7 (#192)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| Fiction.LiveBench | 52.8% | 58.3% |
| LMArena Longer Query | 1422 | 1439 |
Writing & Preference DeepSeek-V3.1 leads
DeepSeek-V3.1: 60.3 (#98), Grok 3: 55.8 (#141)
| Benchmark | DeepSeek-V3.1 | Grok 3 |
|---|---|---|
| LMArena Text | 1420 | 1426 |
| LMArena Creative Writing | 1401 | 1414 |
| EQ-Bench Creative Writing | 1436 | 1186 |
| LMArena Multi-Turn | 1408 | 1425 |
| Short-Story Creative Writing | — | 76.4% |
| WildBench | — | 84.9% |
Frequently asked questions
Is DeepSeek-V3.1 better than Grok 3?
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 39.9 on the Noometry Index.
Is DeepSeek-V3.1 or Grok 3 better for coding?
Grok 3 scores higher on coding benchmarks: 41.9 versus 40.3 in the Noometry coding category.
How many benchmarks do DeepSeek-V3.1 and Grok 3 share?
24 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Grok 3 has 40.