# DeepSeek LLM 67B vs Grok 4

> Grok 4 is the stronger model overall, scoring 48.1 to 24.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-llm-67b-vs-grok-4
- Last updated: 2026-10-10
- Shared benchmarks: 14

## Summary

- They share 14 benchmarks with published results for both. DeepSeek LLM 67B scores higher in 0 categories and Grok 4 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Grok 4 leads 53.8 to 7.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 0.8% for DeepSeek LLM 67B and 84% for Grok 4.
- DeepSeek LLM 67B has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 24.9 | 48.1 |
| Rank | 347 | 56 |
| Context | — | — |
| Input $/M | — | — |
| Output $/M | — | — |
| Weights | Open | Proprietary |

## Coding

- DeepSeek LLM 67B: 31.9 (#278)
- Grok 4: 50.3 (#46)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| LMArena Coding | 1096 | 1408 |
| Aider Polyglot | — | 79.6% |
| WeirdML | — | 45.7% |

## Agentic & Tool Use

- DeepSeek LLM 67B: —
- Grok 4: 32.3 (#68)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| Terminal-Bench | — | 27.2% |
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 21.1% |
| Cybench | — | 43% |
| DeepResearch Bench | — | 47.3% |
| BALROG | — | 43.6% |
| LMArena Search | — | 1142 |
| METR Time Horizons | — | 66.6% |

## Reasoning

- DeepSeek LLM 67B: 16.5 (#304)
- Grok 4: 36.7 (#65)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| Chess Puzzles | 0% | 28% |
| LMArena Hard Prompts | 1070 | 1409 |
| Epoch Capabilities Index | 110.5 | 146.44 |
| ARC-AGI-2 | — | 16% |
| SimpleBench | — | 60.5% |
| Kagi LLM Benchmark | — | 73.6% |
| ARC-AGI-1 | — | 66.7% |
| ForecastBench | — | 60.9 |

## Math

- DeepSeek LLM 67B: 8.7 (#324)
- Grok 4: 48.4 (#64)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 0.8% | 84% |
| LMArena Math | 1108 | 1422 |
| Omni-MATH | — | 60.3% |
| MATH Level 5 | 6.4% | — |
| FrontierMath (Feb 2025 set) | — | 19.7% |
| FrontierMath Tier 4 (v1) | — | 2.1% |

## Knowledge

- DeepSeek LLM 67B: 7.0 (#313)
- Grok 4: 53.8 (#55)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| GPQA Diamond | 24.6% | 87% |
| MMLU-Pro | — | 85.1% |
| Confabulations | — | 12.4% |
| GPQA (HELM) | — | 72.7% |
| LMArena Expert | — | 1415 |

## Multimodal

- DeepSeek LLM 67B: —
- Grok 4: 33.7 (#94)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| LMArena Vision | — | 1210 |
| GeoBench | — | 45% |

## Multilingual

- DeepSeek LLM 67B: 29.4 (#267)
- Grok 4: 51.8 (#103)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| LMArena Non-English | 1073 | 1403 |
| LMArena Chinese | 1132 | 1427 |
| LMArena French | — | 1418 |
| LMArena German | — | 1429 |
| LMArena Japanese | — | 1394 |
| LMArena Korean | — | 1377 |
| LMArena Russian | — | 1410 |
| LMArena Spanish | — | 1420 |

## Instruction Following

- DeepSeek LLM 67B: 55.4 (#277)
- Grok 4: 79.2 (#5)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| LMArena Instruction Following | 1079 | 1387 |
| IFEval | — | 94.9% |

## Long Context

- DeepSeek LLM 67B: 33.1 (#265)
- Grok 4: 63.1 (#4)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| LMArena Longer Query | 1092 | 1409 |
| Fiction.LiveBench | — | 94.4% |

## Writing & Preference

- DeepSeek LLM 67B: 31.6 (#282)
- Grok 4: 58.5 (#116)

| Benchmark | DeepSeek LLM 67B | Grok 4 |
|---|---|---|
| LMArena Text | 1105 | 1411 |
| LMArena Creative Writing | 1067 | 1397 |
| LMArena Multi-Turn | 1082 | 1416 |
| Short-Story Creative Writing | — | 76.9% |
| WildBench | — | 79.7% |

## FAQ

### Is DeepSeek LLM 67B better than Grok 4?

Grok 4 is the stronger model overall, scoring 48.1 to 24.9 on the Noometry Index.

### Is DeepSeek LLM 67B or Grok 4 better for coding?

Grok 4 scores higher on coding benchmarks: 50.3 versus 31.9 in the Noometry coding category.

### How many benchmarks do DeepSeek LLM 67B and Grok 4 share?

14 benchmarks have published results for both models. DeepSeek LLM 67B has 15 scored results on Noometry and Grok 4 has 48.