# Phi 3 Mini 4k Instruct vs Phi-4

> Phi-4 is the stronger model overall, scoring 30.9 to 27.4 on the Noometry Index.

- Canonical page: https://noometry.com/compare/phi-3-mini-4k-instruct-vs-phi-4
- Last updated: 2026-10-10
- Shared benchmarks: 26

## Summary

- They share 26 benchmarks with published results for both. Phi 3 Mini 4k Instruct scores higher in 1 category and Phi-4 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Phi-4 leads 40.7 to 27.7.
- The biggest single-benchmark swing is LiveBench Math: 15.7% for Phi 3 Mini 4k Instruct and 42% for Phi-4.

## Snapshot

| | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| Provider | Microsoft | Microsoft |
| Noometry Index | 27.4 | 30.9 |
| Rank | 324 | 269 |
| Context | — | 128K |
| Input $/M | — | $0.07 |
| Output $/M | — | $0.14 |
| Weights | Open | Open |

## Coding

- Phi 3 Mini 4k Instruct: 26.1 (#319)
- Phi-4: 33.8 (#237)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| LiveBench Coding | 15.5% | 30.7% |
| LMArena Coding | 1093 | 1231 |
| BigCodeBench Instruct | — | 45.5% |
| BigCodeBench Complete | — | 55.4% |
| HumanEval+ | 59.1% | — |
| MBPP+ | 54.2% | — |

## Agentic & Tool Use

- Phi 3 Mini 4k Instruct: —
- Phi-4: 22.6 (#120)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |

## Reasoning

- Phi 3 Mini 4k Instruct: 13.6 (#326)
- Phi-4: 17.0 (#287)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| Chess Puzzles | 0% | 1% |
| LiveBench Reasoning | 26.8% | 47.8% |
| LMArena Hard Prompts | 1072 | 1220 |
| LiveBench Data Analysis | 34.7% | 45.2% |
| LiveBench | 22.4% | 41.6% |
| Adversarial NLI | 52.8% | — |
| BIG-Bench Hard | 71.7% | — |
| Epoch Capabilities Index | — | 130.43 |
| HellaSwag | 76.7% | — |
| WinoGrande | 70.8% | — |

## Math

- Phi 3 Mini 4k Instruct: 24.5 (#258)
- Phi-4: 19.6 (#279)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| LiveBench Math | 15.7% | 42% |
| LMArena Math | 1111 | 1246 |
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| MATH Level 5 | — | 64.9% |

## Knowledge

- Phi 3 Mini 4k Instruct: 27.9 (#241)
- Phi-4: 32.5 (#204)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| LMArena Expert | 1045 | 1203 |
| MMLU | 68.8% | 84.8% |
| GPQA Diamond | — | 56.1% |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| ARC (AI2) Challenge | 84.9% | — |
| OpenBookQA | 88% | — |
| TriviaQA | 64% | — |

## Multilingual

- Phi 3 Mini 4k Instruct: 26.5 (#275)
- Phi-4: 37.4 (#232)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| LMArena Non-English | 1021 | 1197 |
| LMArena Chinese | 1021 | 1211 |
| LMArena French | 1075 | 1223 |
| LMArena German | 1044 | 1222 |
| LMArena Japanese | 935 | 1158 |
| LMArena Korean | 905 | 1151 |
| LMArena Russian | 1022 | 1209 |
| LMArena Spanish | 1084 | 1234 |

## Instruction Following

- Phi 3 Mini 4k Instruct: 48.0 (#298)
- Phi-4: 60.6 (#246)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| LiveBench Instruction Following | 39.1% | 58.4% |
| LMArena Instruction Following | 1053 | 1201 |

## Long Context

- Phi 3 Mini 4k Instruct: 31.7 (#271)
- Phi-4: 36.9 (#221)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1044 | 1217 |

## Writing & Preference

- Phi 3 Mini 4k Instruct: 27.7 (#295)
- Phi-4: 40.7 (#239)

| Benchmark | Phi 3 Mini 4k Instruct | Phi-4 |
|---|---|---|
| LMArena Text | 1073 | 1217 |
| LMArena Creative Writing | 1037 | 1182 |
| LMArena Multi-Turn | 1018 | 1206 |
| LiveBench Language | 9.2% | 25.6% |
| Short-Story Creative Writing | — | 62.6% |

## FAQ

### Is Phi 3 Mini 4k Instruct better than Phi-4?

Phi-4 is the stronger model overall, scoring 30.9 to 27.4 on the Noometry Index.

### Is Phi 3 Mini 4k Instruct or Phi-4 better for coding?

Phi-4 scores higher on coding benchmarks: 33.8 versus 26.1 in the Noometry coding category.

### How many benchmarks do Phi 3 Mini 4k Instruct and Phi-4 share?

26 benchmarks have published results for both models. Phi 3 Mini 4k Instruct has 35 scored results on Noometry and Phi-4 has 37.