# Claude 3 Haiku vs Phi-4

> Phi-4 is the stronger model overall, scoring 31.2 to 25.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-3-haiku-vs-phi-4
- Last updated: 2026-10-11
- Shared benchmarks: 25

## Summary

- They share 25 benchmarks with published results for both. Claude 3 Haiku scores higher in 1 category and Phi-4 in 7 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Phi-4 leads 32.6 to 17.3.
- The biggest single-benchmark swing is MATH Level 5: 14.9% for Claude 3 Haiku and 64.9% for Phi-4.
- Phi-4 has downloadable open weights; the other is API-only.

## Snapshot

| | Claude 3 Haiku | Phi-4 |
|---|---|---|
| Provider | Anthropic | Microsoft |
| Noometry Index | 25.9 | 31.2 |
| Rank | 340 | 279 |
| Context | — | 128K |
| Input $/M | — | $0.07 |
| Output $/M | — | $0.14 |
| Weights | Proprietary | Open |

## Coding

- Claude 3 Haiku: 26.4 (#325)
- Phi-4: 34.4 (#239)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| BigCodeBench Instruct | 39.4% | 45.5% |
| LMArena Coding | 1199 | 1231 |
| BigCodeBench Complete | 50.1% | 55.4% |
| WeirdML | 9.8% | — |
| LiveBench Coding | — | 30.7% |
| CadEval | 12% | — |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |

## Agentic & Tool Use

- Claude 3 Haiku: —
- Phi-4: 22.8 (#128)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |

## Reasoning

- Claude 3 Haiku: 16.3 (#307)
- Phi-4: 17.7 (#291)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1174 | 1220 |
| Epoch Capabilities Index | 118.35 | 130.42 |
| Kagi LLM Benchmark | 34.2% | — |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| DTBench | 50.1% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 8.8% | — |
| ForecastBench | 53.2 | — |
| LiveBench | — | 41.6% |
| WinoGrande | 74.2% | — |

## Math

- Claude 3 Haiku: 9.8 (#319)
- Phi-4: 20.8 (#285)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | 13.8% |
| LMArena Math | 1188 | 1246 |
| MATH Level 5 | 14.9% | 64.9% |
| LiveBench Math | — | 42% |

## Knowledge

- Claude 3 Haiku: 17.3 (#285)
- Phi-4: 32.6 (#209)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| GPQA Diamond | 36.3% | 56.1% |
| Confabulations | 34.2% | 29.4% |
| LMArena Expert | 1148 | 1203 |
| MMLU | 73.8% | 84.8% |
| Vectara Hallucination Rate | — | 3.7% |

## Multimodal

- Claude 3 Haiku: 23.6 (#128)
- Phi-4: —

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| LMArena Vision | 950 | — |
| ScienceQA | 72% | — |

## Multilingual

- Claude 3 Haiku: 36.0 (#243)
- Phi-4: 37.2 (#237)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| LMArena Non-English | 1178 | 1197 |
| LMArena Chinese | 1155 | 1212 |
| LMArena French | 1195 | 1224 |
| LMArena German | 1174 | 1222 |
| LMArena Japanese | 1102 | 1158 |
| LMArena Korean | 1109 | 1151 |
| LMArena Russian | 1204 | 1209 |
| LMArena Spanish | 1166 | 1234 |

## Instruction Following

- Claude 3 Haiku: 61.3 (#247)
- Phi-4: 60.4 (#251)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1173 | 1201 |
| LiveBench Instruction Following | — | 58.4% |

## Long Context

- Claude 3 Haiku: 36.1 (#237)
- Phi-4: 36.9 (#226)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1190 | 1217 |

## Writing & Preference

- Claude 3 Haiku: 29.7 (#291)
- Phi-4: 40.5 (#244)

| Benchmark | Claude 3 Haiku | Phi-4 |
|---|---|---|
| LMArena Text | 1195 | 1217 |
| LMArena Creative Writing | 1157 | 1182 |
| LMArena Multi-Turn | 1190 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench Creative Writing | 717 | — |
| LiveBench Language | — | 25.6% |

## FAQ

### Is Claude 3 Haiku better than Phi-4?

Phi-4 is the stronger model overall, scoring 31.2 to 25.9 on the Noometry Index.

### Is Claude 3 Haiku or Phi-4 better for coding?

Phi-4 scores higher on coding benchmarks: 34.4 versus 26.4 in the Noometry coding category.

### How many benchmarks do Claude 3 Haiku and Phi-4 share?

25 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and Phi-4 has 37.