# DeepSeek V4 Pro vs Grok 3

> DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 39.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-pro-vs-grok-3
- Last updated: 2026-10-10
- Shared benchmarks: 26

## Summary

- They share 26 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and Grok 3 in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.5 to 13.7.
- The biggest single-benchmark swing is ARC-AGI-1: 90.5% for DeepSeek V4 Pro and 5.5% for Grok 3.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 54.3 | 39.9 |
| Rank | 31 | 157 |
| Context | 1M | — |
| Input $/M | $0.66 | — |
| Output $/M | $1.98 | — |
| Weights | Open | Proprietary |

## Coding

- DeepSeek V4 Pro: 52.4 (#34)
- Grok 3: 41.9 (#115)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| WeirdML | 66.2% | 37.2% |
| LMArena Coding | 1470 | 1432 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| Aider Polyglot | — | 53.3% |
| LMArena WebDev | 1582 | — |
| SciCode | 51% | — |
| ALE-Bench | 1,403 | — |

## Agentic & Tool Use

- DeepSeek V4 Pro: 32.8 (#58)
- Grok 3: 30.5 (#76)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| APEX-Agents | 47.3% | — |
| BALROG | — | 29.5% |
| Vending-Bench 2 | 3,285 | — |

## Reasoning

- DeepSeek V4 Pro: 56.5 (#24)
- Grok 3: 13.7 (#333)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| ARC-AGI-2 | 61.3% | 0% |
| Kagi LLM Benchmark | 53.5% | 61.3% |
| ARC-AGI-1 | 90.5% | 5.5% |
| LMArena Hard Prompts | 1461 | 1434 |
| Epoch Capabilities Index | 155.31 | 138.33 |
| SimpleBench | — | 36.1% |
| NYT Connections (extended) | 91.3% | — |
| CritPt | 18% | — |
| Chess Puzzles | 47% | — |
| Mystery Game Puzzles | 43% | — |
| DTBench | 93.9% | — |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |

## Math

- DeepSeek V4 Pro: 64.8 (#30)
- Grok 3: 38.0 (#145)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.6% | 55.6% |
| LMArena Math | 1455 | 1391 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| Omni-MATH | — | 46.4% |
| MATH Level 5 | — | 88.7% |
| FrontierMath (Feb 2025 set) | — | 3.8% |
| FrontierMath Tier 4 (v1) | — | 0% |

## Knowledge

- DeepSeek V4 Pro: 59.5 (#31)
- Grok 3: 46.2 (#82)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| GPQA Diamond | 91.7% | 75.8% |
| Vectara Hallucination Rate | 8.6% | 5.8% |
| LMArena Expert | 1464 | 1421 |
| SimpleQA Verified | 52.9% | — |
| MMLU-Pro | — | 78.8% |
| Confabulations | — | 14.2% |
| GPQA (HELM) | — | 65% |

## Multilingual

- DeepSeek V4 Pro: 54.4 (#45)
- Grok 3: 52.3 (#87)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| LMArena Non-English | 1439 | 1410 |
| LMArena Chinese | 1486 | 1448 |
| LMArena French | 1472 | 1460 |
| LMArena German | 1458 | 1431 |
| LMArena Japanese | 1445 | 1387 |
| LMArena Korean | 1447 | 1373 |
| LMArena Russian | 1453 | 1416 |
| LMArena Spanish | 1458 | 1417 |

## Instruction Following

- DeepSeek V4 Pro: 76.1 (#47)
- Grok 3: 75.0 (#73)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| LMArena Instruction Following | 1448 | 1409 |
| IFEval | — | 88.4% |

## Long Context

- DeepSeek V4 Pro: 45.0 (#51)
- Grok 3: 38.7 (#192)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| LMArena Longer Query | 1458 | 1439 |
| Fiction.LiveBench | — | 58.3% |
| CL-bench Life | 13.5% | — |

## Writing & Preference

- DeepSeek V4 Pro: 65.5 (#46)
- Grok 3: 55.8 (#141)

| Benchmark | DeepSeek V4 Pro | Grok 3 |
|---|---|---|
| LMArena Text | 1451 | 1426 |
| LMArena Creative Writing | 1446 | 1414 |
| EQ-Bench Creative Writing | 1553 | 1186 |
| LMArena Multi-Turn | 1467 | 1425 |
| Short-Story Creative Writing | — | 76.4% |
| WildBench | — | 84.9% |
| EQ-Bench 4 | 1166 | — |

## FAQ

### Is DeepSeek V4 Pro better than Grok 3?

DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 39.9 on the Noometry Index.

### Is DeepSeek V4 Pro or Grok 3 better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 41.9 in the Noometry coding category.

### How many benchmarks do DeepSeek V4 Pro and Grok 3 share?

26 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Grok 3 has 40.