# Mixtral 8x22B vs Phi-4

> Phi-4 is the stronger model overall, scoring 31.2 to 27.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/mixtral-8x22b-vs-phi-4
- Last updated: 2026-10-12
- Shared benchmarks: 23

## Summary

- They share 23 benchmarks with published results for both. Mixtral 8x22B scores higher in 3 categories and Phi-4 in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Phi-4 leads 32.6 to 15.1.
- The biggest single-benchmark swing is MATH Level 5: 24.2% for Mixtral 8x22B and 64.9% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $2 / $6 for Mixtral 8x22B.
- Phi-4 accepts more context: 128K tokens versus 64K.

## Snapshot

| | Mixtral 8x22B | Phi-4 |
|---|---|---|
| Provider | Mistral AI | Microsoft |
| Noometry Index | 27.1 | 31.2 |
| Rank | 333 | 279 |
| Context | 64K | 128K |
| Input $/M | $2 | $0.07 |
| Output $/M | $6 | $0.14 |
| Weights | Open | Open |

## Coding

- Mixtral 8x22B: 24.2 (#329)
- Phi-4: 34.4 (#239)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| BigCodeBench Instruct | 40.6% | 45.5% |
| LMArena Coding | 1166 | 1231 |
| BigCodeBench Complete | 50.2% | 55.4% |
| WeirdML | 3.2% | — |
| LiveBench Coding | — | 30.7% |
| HumanEval+ | 72% | — |
| MBPP+ | 64.3% | — |

## Agentic & Tool Use

- Mixtral 8x22B: 23.1 (#127)
- Phi-4: 22.8 (#128)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| Cybench | 7.5% | — |
| BALROG | — | 11.6% |

## Reasoning

- Mixtral 8x22B: 19.9 (#248)
- Phi-4: 17.7 (#291)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1150 | 1220 |
| Epoch Capabilities Index | 122.03 | 130.42 |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| DTBench | 55.1% | — |
| LiveBench Data Analysis | — | 45.2% |
| ForecastBench | 56.3 | — |
| LiveBench | — | 41.6% |

## Math

- Mixtral 8x22B: 22.9 (#275)
- Phi-4: 20.8 (#285)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| LMArena Math | 1184 | 1246 |
| MATH Level 5 | 24.2% | 64.9% |
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| Omni-MATH | 16.3% | — |
| LiveBench Math | — | 42% |

## Knowledge

- Mixtral 8x22B: 15.1 (#293)
- Phi-4: 32.6 (#209)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| GPQA Diamond | 34.1% | 56.1% |
| LMArena Expert | 1113 | 1203 |
| MMLU | 77.8% | 84.8% |
| MMLU-Pro | 46% | — |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| GPQA (HELM) | 33.4% | — |

## Multilingual

- Mixtral 8x22B: 32.8 (#255)
- Phi-4: 37.2 (#237)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| LMArena Non-English | 1128 | 1197 |
| LMArena Chinese | 1116 | 1212 |
| LMArena French | 1166 | 1224 |
| LMArena German | 1141 | 1222 |
| LMArena Japanese | 1037 | 1158 |
| LMArena Korean | 1057 | 1151 |
| LMArena Russian | 1158 | 1209 |
| LMArena Spanish | 1151 | 1234 |

## Instruction Following

- Mixtral 8x22B: 57.7 (#266)
- Phi-4: 60.4 (#251)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1147 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
| IFEval | 72.4% | — |

## Long Context

- Mixtral 8x22B: 34.7 (#247)
- Phi-4: 36.9 (#226)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1144 | 1217 |

## Writing & Preference

- Mixtral 8x22B: 36.9 (#262)
- Phi-4: 40.5 (#244)

| Benchmark | Mixtral 8x22B | Phi-4 |
|---|---|---|
| LMArena Text | 1162 | 1217 |
| LMArena Creative Writing | 1141 | 1182 |
| LMArena Multi-Turn | 1130 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| WildBench | 71.1% | — |
| LiveBench Language | — | 25.6% |

## FAQ

### Is Mixtral 8x22B better than Phi-4?

Phi-4 is the stronger model overall, scoring 31.2 to 27.1 on the Noometry Index.

### Which is cheaper, Mixtral 8x22B or Phi-4?

Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Mixtral 8x22B lists at $2 and $6.

### Is Mixtral 8x22B or Phi-4 better for coding?

Phi-4 scores higher on coding benchmarks: 34.4 versus 24.2 in the Noometry coding category.

### Which has the bigger context window?

Phi-4 does, with 128K tokens against 64K.

### How many benchmarks do Mixtral 8x22B and Phi-4 share?

23 benchmarks have published results for both models. Mixtral 8x22B has 34 scored results on Noometry and Phi-4 has 37.