# DeepSeek-V3.2-Speciale vs Grok 3

> DeepSeek-V3.2-Speciale and Grok 3 score almost the same on the Noometry Index (39.7 vs 39.9), so choose on price, context window or the category you care about most.

- Canonical page: https://noometry.com/compare/deepseek-v3-2-speciale-vs-grok-3
- Last updated: 2026-10-10
- Shared benchmarks: 3

## Summary

- They share 3 benchmarks with published results for both. DeepSeek-V3.2-Speciale scores higher in 1 category and Grok 3 in 2 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek-V3.2-Speciale leads 32.9 to 13.7.
- The biggest single-benchmark swing is SimpleBench: 52.6% for DeepSeek-V3.2-Speciale and 36.1% for Grok 3.
- DeepSeek-V3.2-Speciale has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 39.7 | 39.9 |
| Rank | 162 | 157 |
| Context | 128K | — |
| Input $/M | $0.58 | — |
| Output $/M | $1.68 | — |
| Weights | Open | Proprietary |

## Coding

- DeepSeek-V3.2-Speciale: 40.4 (#140)
- Grok 3: 41.9 (#115)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| WeirdML | 46.7% | 37.2% |
| Aider Polyglot | — | 53.3% |
| LMArena Coding | — | 1432 |

## Agentic & Tool Use

- DeepSeek-V3.2-Speciale: —
- Grok 3: 30.5 (#76)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| BALROG | — | 29.5% |

## Reasoning

- DeepSeek-V3.2-Speciale: 32.9 (#73)
- Grok 3: 13.7 (#333)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| SimpleBench | 52.6% | 36.1% |
| ARC-AGI-2 | — | 0% |
| Kagi LLM Benchmark | — | 61.3% |
| ARC-AGI-1 | — | 5.5% |
| LMArena Hard Prompts | — | 1434 |
| Epoch Capabilities Index | — | 138.33 |

## Math

- DeepSeek-V3.2-Speciale: —
- Grok 3: 38.0 (#145)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 55.6% |
| Omni-MATH | — | 46.4% |
| LMArena Math | — | 1391 |
| MATH Level 5 | — | 88.7% |
| FrontierMath (Feb 2025 set) | — | 3.8% |
| FrontierMath Tier 4 (v1) | — | 0% |

## Knowledge

- DeepSeek-V3.2-Speciale: —
- Grok 3: 46.2 (#82)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| GPQA Diamond | — | 75.8% |
| MMLU-Pro | — | 78.8% |
| Confabulations | — | 14.2% |
| Vectara Hallucination Rate | — | 5.8% |
| GPQA (HELM) | — | 65% |
| LMArena Expert | — | 1421 |

## Multilingual

- DeepSeek-V3.2-Speciale: —
- Grok 3: 52.3 (#87)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| LMArena Non-English | — | 1410 |
| LMArena Chinese | — | 1448 |
| LMArena French | — | 1460 |
| LMArena German | — | 1431 |
| LMArena Japanese | — | 1387 |
| LMArena Korean | — | 1373 |
| LMArena Russian | — | 1416 |
| LMArena Spanish | — | 1417 |

## Instruction Following

- DeepSeek-V3.2-Speciale: —
- Grok 3: 75.0 (#73)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| IFEval | — | 88.4% |
| LMArena Instruction Following | — | 1409 |

## Long Context

- DeepSeek-V3.2-Speciale: —
- Grok 3: 38.7 (#192)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| Fiction.LiveBench | — | 58.3% |
| LMArena Longer Query | — | 1439 |

## Writing & Preference

- DeepSeek-V3.2-Speciale: 46.0 (#222)
- Grok 3: 55.8 (#141)

| Benchmark | DeepSeek-V3.2-Speciale | Grok 3 |
|---|---|---|
| EQ-Bench Creative Writing | 1276 | 1186 |
| LMArena Text | — | 1426 |
| LMArena Creative Writing | — | 1414 |
| Short-Story Creative Writing | — | 76.4% |
| WildBench | — | 84.9% |
| LMArena Multi-Turn | — | 1425 |

## FAQ

### Is DeepSeek-V3.2-Speciale better than Grok 3?

DeepSeek-V3.2-Speciale and Grok 3 score almost the same on the Noometry Index (39.7 vs 39.9), so choose on price, context window or the category you care about most.

### Is DeepSeek-V3.2-Speciale or Grok 3 better for coding?

Grok 3 scores higher on coding benchmarks: 41.9 versus 40.4 in the Noometry coding category.

### How many benchmarks do DeepSeek-V3.2-Speciale and Grok 3 share?

3 benchmarks have published results for both models. DeepSeek-V3.2-Speciale has 3 scored results on Noometry and Grok 3 has 40.