# Deepseek Coder v2 vs Phi 3 Mini 4k Instruct

> Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 27.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-coder-v2-vs-phi-3-mini-4k-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 21

## Summary

- They share 21 benchmarks with published results for both. Deepseek Coder v2 scores higher in 8 categories and Phi 3 Mini 4k Instruct in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in instruction following, where Deepseek Coder v2 leads 61.7 to 47.7.

## Snapshot

| | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| Provider | DeepSeek | Microsoft |
| Noometry Index | 35.9 | 27.9 |
| Rank | 220 | 328 |
| Context | — | — |
| Input $/M | — | — |
| Output $/M | — | — |
| Weights | Open | Open |

## Coding

- Deepseek Coder v2: 38.1 (#183)
- Phi 3 Mini 4k Instruct: 26.6 (#323)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Coding | 1251 | 1093 |
| HumanEval+ | 82.3% | 59.1% |
| MBPP+ | 75.1% | 54.2% |
| BigCodeBench Instruct | 48.2% | — |
| LiveBench Coding | — | 15.5% |
| BigCodeBench Complete | 59.7% | — |

## Reasoning

- Deepseek Coder v2: 23.6 (#176)
- Phi 3 Mini 4k Instruct: 14.1 (#328)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Hard Prompts | 1207 | 1072 |
| WinoGrande | 83.7% | 70.8% |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | — | 26.8% |
| LiveBench Data Analysis | — | 34.7% |
| Adversarial NLI | — | 52.8% |
| BIG-Bench Hard | — | 71.7% |
| HellaSwag | — | 76.7% |
| LiveBench | — | 22.4% |

## Math

- Deepseek Coder v2: 34.9 (#190)
- Phi 3 Mini 4k Instruct: 26.6 (#257)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Math | 1241 | 1111 |
| LiveBench Math | — | 15.7% |
| GSM8K | 94.5% | — |

## Knowledge

- Deepseek Coder v2: 32.3 (#212)
- Phi 3 Mini 4k Instruct: 28.5 (#246)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Expert | 1181 | 1045 |
| ARC (AI2) Challenge | 64.3% | 84.9% |
| MMLU | — | 68.8% |
| OpenBookQA | — | 88% |
| TriviaQA | — | 64% |

## Multilingual

- Deepseek Coder v2: 36.3 (#240)
- Phi 3 Mini 4k Instruct: 26.3 (#280)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Non-English | 1182 | 1021 |
| LMArena Chinese | 1201 | 1021 |
| LMArena French | 1185 | 1076 |
| LMArena German | 1164 | 1044 |
| LMArena Japanese | 1126 | 935 |
| LMArena Korean | 1104 | 905 |
| LMArena Russian | 1188 | 1022 |
| LMArena Spanish | 1153 | 1085 |

## Instruction Following

- Deepseek Coder v2: 61.7 (#242)
- Phi 3 Mini 4k Instruct: 47.7 (#303)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Instruction Following | 1180 | 1053 |
| LiveBench Instruction Following | — | 39.1% |

## Long Context

- Deepseek Coder v2: 37.0 (#224)
- Phi 3 Mini 4k Instruct: 31.7 (#276)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Longer Query | 1219 | 1044 |

## Writing & Preference

- Deepseek Coder v2: 38.2 (#253)
- Phi 3 Mini 4k Instruct: 27.6 (#300)

| Benchmark | Deepseek Coder v2 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Text | 1191 | 1073 |
| LMArena Creative Writing | 1120 | 1037 |
| LMArena Multi-Turn | 1177 | 1018 |
| LiveBench Language | — | 9.2% |

## FAQ

### Is Deepseek Coder v2 better than Phi 3 Mini 4k Instruct?

Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 27.9 on the Noometry Index.

### Is Deepseek Coder v2 or Phi 3 Mini 4k Instruct better for coding?

Deepseek Coder v2 scores higher on coding benchmarks: 38.1 versus 26.6 in the Noometry coding category.

### How many benchmarks do Deepseek Coder v2 and Phi 3 Mini 4k Instruct share?

21 benchmarks have published results for both models. Deepseek Coder v2 has 24 scored results on Noometry and Phi 3 Mini 4k Instruct has 35.