# C4ai Aya Expanse 32b vs Phi-4

> C4ai Aya Expanse 32b is the stronger model overall, scoring 35.3 to 30.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/c4ai-aya-expanse-32b-vs-phi-4
- Last updated: 2026-10-10
- Shared benchmarks: 18

## Summary

- They share 18 benchmarks with published results for both. C4ai Aya Expanse 32b scores higher in 8 categories and Phi-4 in 0 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in math, where C4ai Aya Expanse 32b leads 31.4 to 19.6.
- The biggest single-benchmark swing is Vectara Hallucination Rate: 10.9% for C4ai Aya Expanse 32b and 3.7% for Phi-4.

## Snapshot

| | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| Provider | Cohere | Microsoft |
| Noometry Index | 35.3 | 30.9 |
| Rank | 216 | 269 |
| Context | 128K | 128K |
| Input $/M | — | $0.07 |
| Output $/M | — | $0.14 |
| Weights | Open | Open |

## Coding

- C4ai Aya Expanse 32b: 34.3 (#227)
- Phi-4: 33.8 (#237)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| LMArena Coding | 1197 | 1231 |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |

## Agentic & Tool Use

- C4ai Aya Expanse 32b: —
- Phi-4: 22.6 (#120)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |

## Reasoning

- C4ai Aya Expanse 32b: 22.3 (#183)
- Phi-4: 17.0 (#287)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1193 | 1220 |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| Epoch Capabilities Index | — | 130.43 |
| LiveBench | — | 41.6% |

## Math

- C4ai Aya Expanse 32b: 31.4 (#195)
- Phi-4: 19.6 (#279)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| LMArena Math | 1200 | 1246 |
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |

## Knowledge

- C4ai Aya Expanse 32b: 32.6 (#202)
- Phi-4: 32.5 (#204)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| Vectara Hallucination Rate | 10.9% | 3.7% |
| LMArena Expert | 1182 | 1203 |
| GPQA Diamond | — | 56.1% |
| Confabulations | — | 29.4% |
| MMLU | — | 84.8% |

## Multilingual

- C4ai Aya Expanse 32b: 38.5 (#225)
- Phi-4: 37.4 (#232)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| LMArena Non-English | 1213 | 1197 |
| LMArena Chinese | 1210 | 1211 |
| LMArena French | 1247 | 1223 |
| LMArena German | 1199 | 1222 |
| LMArena Japanese | 1163 | 1158 |
| LMArena Korean | 1158 | 1151 |
| LMArena Russian | 1227 | 1209 |
| LMArena Spanish | 1193 | 1234 |

## Instruction Following

- C4ai Aya Expanse 32b: 62.9 (#232)
- Phi-4: 60.6 (#246)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1196 | 1201 |
| LiveBench Instruction Following | — | 58.4% |

## Long Context

- C4ai Aya Expanse 32b: 37.2 (#215)
- Phi-4: 36.9 (#221)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1228 | 1217 |

## Writing & Preference

- C4ai Aya Expanse 32b: 42.4 (#230)
- Phi-4: 40.7 (#239)

| Benchmark | C4ai Aya Expanse 32b | Phi-4 |
|---|---|---|
| LMArena Text | 1224 | 1217 |
| LMArena Creative Writing | 1200 | 1182 |
| LMArena Multi-Turn | 1190 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| LiveBench Language | — | 25.6% |

## FAQ

### Is C4ai Aya Expanse 32b better than Phi-4?

C4ai Aya Expanse 32b is the stronger model overall, scoring 35.3 to 30.9 on the Noometry Index.

### Is C4ai Aya Expanse 32b or Phi-4 better for coding?

They score almost the same on coding (34.3 vs 33.8); test both on your own repository before choosing.

### Which has the bigger context window?

Both accept 128K tokens.

### How many benchmarks do C4ai Aya Expanse 32b and Phi-4 share?

18 benchmarks have published results for both models. C4ai Aya Expanse 32b has 18 scored results on Noometry and Phi-4 has 37.