# Gemma 3 4B vs Grok-2 (Dec 2024)

> Grok-2 (Dec 2024) is the stronger model overall, scoring 33.7 to 28.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gemma-3-4b-vs-grok-2
- Last updated: 2026-10-12
- Shared benchmarks: 16

## Summary

- They share 16 benchmarks with published results for both. Gemma 3 4B scores higher in 1 category and Grok-2 (Dec 2024) in 7 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Grok-2 (Dec 2024) leads 29.8 to 11.8.
- The biggest single-benchmark swing is GPQA Diamond: 23.2% for Gemma 3 4B and 53.8% for Grok-2 (Dec 2024).
- Gemma 3 4B has downloadable open weights; the other is API-only.

## Snapshot

| | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| Provider | Google | xAI |
| Noometry Index | 28.1 | 33.7 |
| Rank | 326 | 239 |
| Context | 131K | — |
| Input $/M | $0.04 | — |
| Output $/M | $0.08 | — |
| Weights | Open | Proprietary |

## Coding

- Gemma 3 4B: 35.9 (#215)
- Grok-2 (Dec 2024): 33.3 (#258)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Coding | 1230 | 1287 |
| WeirdML | — | 22.2% |
| LiveBench Coding | — | 46.4% |

## Agentic & Tool Use

- Gemma 3 4B: 20.9 (#142)
- Grok-2 (Dec 2024): —

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | — |

## Reasoning

- Gemma 3 4B: 13.2 (#335)
- Grok-2 (Dec 2024): 16.9 (#299)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Hard Prompts | 1253 | 1272 |
| DTBench | 50.9% | 65.2% |
| Epoch Capabilities Index | 116.02 | 130.48 |
| SimpleBench | — | 22.7% |
| Kagi LLM Benchmark | 25.2% | — |
| Chess Puzzles | 0% | — |
| LiveBench Reasoning | — | 54.8% |
| LiveBench Data Analysis | — | 54.5% |
| LMCA | 2.8% | — |
| LiveBench | — | 54.3% |

## Math

- Gemma 3 4B: 16.8 (#292)
- Grok-2 (Dec 2024): 20.8 (#284)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.5% | 11.5% |
| LMArena Math | 1239 | 1283 |
| LiveBench Math | — | 54.9% |
| MATH Level 5 | — | 63.5% |
| FrontierMath (Feb 2025 set) | — | 0.7% |

## Knowledge

- Gemma 3 4B: 11.8 (#299)
- Grok-2 (Dec 2024): 29.8 (#233)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| GPQA Diamond | 23.2% | 53.8% |
| LMArena Expert | 1223 | 1254 |
| Confabulations | — | 20.1% |
| Vectara Hallucination Rate | 6.4% | — |

## Multilingual

- Gemma 3 4B: 42.5 (#194)
- Grok-2 (Dec 2024): 43.1 (#188)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Non-English | 1273 | 1282 |
| LMArena German | 1281 | 1287 |
| LMArena Russian | 1294 | 1286 |
| LMArena Chinese | — | 1289 |
| LMArena French | — | 1318 |
| LMArena Japanese | — | 1244 |
| LMArena Korean | — | 1237 |
| LMArena Spanish | — | 1281 |

## Instruction Following

- Gemma 3 4B: 65.2 (#225)
- Grok-2 (Dec 2024): 66.9 (#202)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Instruction Following | 1239 | 1270 |
| LiveBench Instruction Following | — | 69.6% |

## Long Context

- Gemma 3 4B: 38.7 (#194)
- Grok-2 (Dec 2024): 38.8 (#190)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Longer Query | 1273 | 1276 |

## Writing & Preference

- Gemma 3 4B: 42.0 (#239)
- Grok-2 (Dec 2024): 48.6 (#198)

| Benchmark | Gemma 3 4B | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Text | 1291 | 1305 |
| LMArena Creative Writing | 1271 | 1284 |
| LMArena Multi-Turn | 1255 | 1290 |
| Short-Story Creative Writing | — | 63.6% |
| EQ-Bench Creative Writing | 1068 | — |
| LiveBench Language | — | 45.6% |

## FAQ

### Is Gemma 3 4B better than Grok-2 (Dec 2024)?

Grok-2 (Dec 2024) is the stronger model overall, scoring 33.7 to 28.1 on the Noometry Index.

### Is Gemma 3 4B or Grok-2 (Dec 2024) better for coding?

Gemma 3 4B scores higher on coding benchmarks: 35.9 versus 33.3 in the Noometry coding category.

### How many benchmarks do Gemma 3 4B and Grok-2 (Dec 2024) share?

16 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and Grok-2 (Dec 2024) has 34.