# Phi-4 vs Sonar

> Sonar is the stronger model overall, scoring 37.8 to 30.9 on the Noometry Index. Phi-4 costs 11× less per token, which makes it the better buy when Sonar's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/phi-4-vs-sonar
- Last updated: 2026-10-10
- Shared benchmarks: 7

## Summary

- They share 7 benchmarks with published results for both. Phi-4 scores higher in 0 categories and Sonar in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Sonar leads 52.6 to 40.7.
- The biggest single-benchmark swing is LiveBench Language: 25.6% for Phi-4 and 44.1% for Sonar.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $1 / $1 for Sonar.
- Phi-4 has downloadable open weights; the other is API-only.

## Snapshot

| | Phi-4 | Sonar |
|---|---|---|
| Provider | Microsoft | Perplexity |
| Noometry Index | 30.9 | 37.8 |
| Rank | 269 | 184 |
| Context | 128K | 128K |
| Input $/M | $0.07 | $1 |
| Output $/M | $0.14 | $1 |
| Weights | Open | Proprietary |

## Coding

- Phi-4: 33.8 (#237)
- Sonar: 35.1 (#217)

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| LiveBench Coding | 30.7% | 35.1% |
| BigCodeBench Instruct | 45.5% | — |
| LMArena Coding | 1231 | — |
| BigCodeBench Complete | 55.4% | — |

## Agentic & Tool Use

- Phi-4: 22.6 (#120)
- Sonar: —

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| Berkeley Function Calling Leaderboard | 28.8% | — |
| BALROG | 11.6% | — |

## Reasoning

- Phi-4: 17.0 (#287)
- Sonar: 20.3 (#221)

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| LiveBench Reasoning | 47.8% | 46.3% |
| LiveBench Data Analysis | 45.2% | 37.9% |
| LiveBench | 41.6% | 46.9% |
| Chess Puzzles | 1% | — |
| LMArena Hard Prompts | 1220 | — |
| Epoch Capabilities Index | 130.43 | — |

## Math

- Phi-4: 19.6 (#279)
- Sonar: 31.3 (#197)

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| LiveBench Math | 42% | 41.6% |
| OTIS Mock AIME 2024-2025 | 13.8% | — |
| LMArena Math | 1246 | — |
| MATH Level 5 | 64.9% | — |

## Knowledge

- Phi-4: 32.5 (#204)
- Sonar: —

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| GPQA Diamond | 56.1% | — |
| Confabulations | 29.4% | — |
| Vectara Hallucination Rate | 3.7% | — |
| LMArena Expert | 1203 | — |
| MMLU | 84.8% | — |

## Multilingual

- Phi-4: 37.4 (#232)
- Sonar: —

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| LMArena Non-English | 1197 | — |
| LMArena Chinese | 1211 | — |
| LMArena French | 1223 | — |
| LMArena German | 1222 | — |
| LMArena Japanese | 1158 | — |
| LMArena Korean | 1151 | — |
| LMArena Russian | 1209 | — |
| LMArena Spanish | 1234 | — |

## Instruction Following

- Phi-4: 60.6 (#246)
- Sonar: 71.5 (#146)

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| LiveBench Instruction Following | 58.4% | 76.2% |
| LMArena Instruction Following | 1201 | — |

## Long Context

- Phi-4: 36.9 (#221)
- Sonar: —

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| LMArena Longer Query | 1217 | — |

## Writing & Preference

- Phi-4: 40.7 (#239)
- Sonar: 52.6 (#166)

| Benchmark | Phi-4 | Sonar |
|---|---|---|
| LiveBench Language | 25.6% | 44.1% |
| LMArena Text | 1217 | — |
| LMArena Creative Writing | 1182 | — |
| Short-Story Creative Writing | 62.6% | — |
| LMArena Multi-Turn | 1206 | — |

## FAQ

### Is Phi-4 better than Sonar?

Sonar is the stronger model overall, scoring 37.8 to 30.9 on the Noometry Index. Phi-4 costs 11× less per token, which makes it the better buy when Sonar's lead doesn't matter for your workload.

### Which is cheaper, Phi-4 or Sonar?

Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Sonar lists at $1 and $1.

### Is Phi-4 or Sonar better for coding?

Sonar scores higher on coding benchmarks: 35.1 versus 33.8 in the Noometry coding category.

### Which has the bigger context window?

Both accept 128K tokens.

### How many benchmarks do Phi-4 and Sonar share?

7 benchmarks have published results for both models. Phi-4 has 37 scored results on Noometry and Sonar has 7.