# Grok-2 (Dec 2024) vs Phi-4

> Grok-2 (Dec 2024) is the stronger model overall, scoring 33.7 to 31.2 on the Noometry Index.

- Canonical page: https://noometry.com/compare/grok-2-vs-phi-4
- Last updated: 2026-10-11
- Shared benchmarks: 30

## Summary

- They share 30 benchmarks with published results for both. Grok-2 (Dec 2024) scores higher in 5 categories and Phi-4 in 3 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Grok-2 (Dec 2024) leads 48.6 to 40.5.
- The biggest single-benchmark swing is LiveBench Language: 45.6% for Grok-2 (Dec 2024) and 25.6% for Phi-4.
- Phi-4 has downloadable open weights; the other is API-only.

## Snapshot

| | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| Provider | xAI | Microsoft |
| Noometry Index | 33.7 | 31.2 |
| Rank | 239 | 279 |
| Context | — | 128K |
| Input $/M | — | $0.07 |
| Output $/M | — | $0.14 |
| Weights | Proprietary | Open |

## Coding

- Grok-2 (Dec 2024): 33.3 (#258)
- Phi-4: 34.4 (#239)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| LiveBench Coding | 46.4% | 30.7% |
| LMArena Coding | 1287 | 1231 |
| WeirdML | 22.2% | — |
| BigCodeBench Instruct | — | 45.5% |
| BigCodeBench Complete | — | 55.4% |

## Agentic & Tool Use

- Grok-2 (Dec 2024): —
- Phi-4: 22.8 (#128)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |

## Reasoning

- Grok-2 (Dec 2024): 16.9 (#299)
- Phi-4: 17.7 (#291)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| LiveBench Reasoning | 54.8% | 47.8% |
| LMArena Hard Prompts | 1272 | 1220 |
| LiveBench Data Analysis | 54.5% | 45.2% |
| Epoch Capabilities Index | 130.48 | 130.42 |
| LiveBench | 54.3% | 41.6% |
| SimpleBench | 22.7% | — |
| Chess Puzzles | — | 1% |
| DTBench | 65.2% | — |

## Math

- Grok-2 (Dec 2024): 20.8 (#284)
- Phi-4: 20.8 (#285)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 11.5% | 13.8% |
| LiveBench Math | 54.9% | 42% |
| LMArena Math | 1283 | 1246 |
| MATH Level 5 | 63.5% | 64.9% |
| FrontierMath (Feb 2025 set) | 0.7% | — |

## Knowledge

- Grok-2 (Dec 2024): 29.8 (#233)
- Phi-4: 32.6 (#209)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| GPQA Diamond | 53.8% | 56.1% |
| Confabulations | 20.1% | 29.4% |
| LMArena Expert | 1254 | 1203 |
| Vectara Hallucination Rate | — | 3.7% |
| MMLU | — | 84.8% |

## Multilingual

- Grok-2 (Dec 2024): 43.1 (#188)
- Phi-4: 37.2 (#237)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| LMArena Non-English | 1282 | 1197 |
| LMArena Chinese | 1289 | 1212 |
| LMArena French | 1318 | 1224 |
| LMArena German | 1287 | 1222 |
| LMArena Japanese | 1244 | 1158 |
| LMArena Korean | 1237 | 1151 |
| LMArena Russian | 1286 | 1209 |
| LMArena Spanish | 1281 | 1234 |

## Instruction Following

- Grok-2 (Dec 2024): 66.9 (#202)
- Phi-4: 60.4 (#251)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| LiveBench Instruction Following | 69.6% | 58.4% |
| LMArena Instruction Following | 1270 | 1201 |

## Long Context

- Grok-2 (Dec 2024): 38.8 (#190)
- Phi-4: 36.9 (#226)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1276 | 1217 |

## Writing & Preference

- Grok-2 (Dec 2024): 48.6 (#198)
- Phi-4: 40.5 (#244)

| Benchmark | Grok-2 (Dec 2024) | Phi-4 |
|---|---|---|
| LMArena Text | 1305 | 1217 |
| LMArena Creative Writing | 1284 | 1182 |
| Short-Story Creative Writing | 63.6% | 62.6% |
| LMArena Multi-Turn | 1290 | 1206 |
| LiveBench Language | 45.6% | 25.6% |

## FAQ

### Is Grok-2 (Dec 2024) better than Phi-4?

Grok-2 (Dec 2024) is the stronger model overall, scoring 33.7 to 31.2 on the Noometry Index.

### Is Grok-2 (Dec 2024) or Phi-4 better for coding?

Phi-4 scores higher on coding benchmarks: 34.4 versus 33.3 in the Noometry coding category.

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

30 benchmarks have published results for both models. Grok-2 (Dec 2024) has 34 scored results on Noometry and Phi-4 has 37.