# Olmo 3 32b Think vs Phi-4

> Olmo 3 32b Think is the stronger model overall, scoring 38.7 to 31.2 on the Noometry Index.

- Canonical page: https://noometry.com/compare/olmo-3-32b-think-vs-phi-4
- Last updated: 2026-10-11
- Shared benchmarks: 14

## Summary

- They share 14 benchmarks with published results for both. Olmo 3 32b Think scores higher in 8 categories and Phi-4 in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Olmo 3 32b Think leads 36.5 to 20.8.

## Snapshot

| | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| Provider | Allen Institute for AI (Ai2) | Microsoft |
| Noometry Index | 38.7 | 31.2 |
| Rank | 183 | 279 |
| Context | — | 128K |
| Input $/M | — | $0.07 |
| Output $/M | — | $0.14 |
| Weights | Open | Open |

## Coding

- Olmo 3 32b Think: 38.6 (#172)
- Phi-4: 34.4 (#239)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Coding | 1319 | 1231 |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |

## Agentic & Tool Use

- Olmo 3 32b Think: —
- Phi-4: 22.8 (#128)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |

## Reasoning

- Olmo 3 32b Think: 25.9 (#140)
- Phi-4: 17.7 (#291)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1302 | 1220 |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| Epoch Capabilities Index | — | 130.42 |
| LiveBench | — | 41.6% |

## Math

- Olmo 3 32b Think: 36.5 (#165)
- Phi-4: 20.8 (#285)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Math | 1316 | 1246 |
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |

## Knowledge

- Olmo 3 32b Think: 35.0 (#190)
- Phi-4: 32.6 (#209)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Expert | 1273 | 1203 |
| GPQA Diamond | — | 56.1% |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| MMLU | — | 84.8% |

## Multilingual

- Olmo 3 32b Think: 41.2 (#210)
- Phi-4: 37.2 (#237)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Non-English | 1255 | 1197 |
| LMArena Chinese | 1300 | 1212 |
| LMArena French | 1291 | 1224 |
| LMArena German | 1290 | 1222 |
| LMArena Russian | 1254 | 1209 |
| LMArena Japanese | — | 1158 |
| LMArena Korean | — | 1151 |
| LMArena Spanish | — | 1234 |

## Instruction Following

- Olmo 3 32b Think: 67.2 (#198)
- Phi-4: 60.4 (#251)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1275 | 1201 |
| LiveBench Instruction Following | — | 58.4% |

## Long Context

- Olmo 3 32b Think: 39.4 (#182)
- Phi-4: 36.9 (#226)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1296 | 1217 |

## Writing & Preference

- Olmo 3 32b Think: 49.1 (#193)
- Phi-4: 40.5 (#244)

| Benchmark | Olmo 3 32b Think | Phi-4 |
|---|---|---|
| LMArena Text | 1300 | 1217 |
| LMArena Creative Writing | 1256 | 1182 |
| LMArena Multi-Turn | 1290 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| LiveBench Language | — | 25.6% |

## FAQ

### Is Olmo 3 32b Think better than Phi-4?

Olmo 3 32b Think is the stronger model overall, scoring 38.7 to 31.2 on the Noometry Index.

### Is Olmo 3 32b Think or Phi-4 better for coding?

Olmo 3 32b Think scores higher on coding benchmarks: 38.6 versus 34.4 in the Noometry coding category.

### How many benchmarks do Olmo 3 32b Think and Phi-4 share?

14 benchmarks have published results for both models. Olmo 3 32b Think has 14 scored results on Noometry and Phi-4 has 37.