# DeepSeek-V3 vs Phi 3 Small 8k Instruct

> DeepSeek-V3 is the stronger model overall, scoring 39.2 to 28.8 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v3-vs-phi-3-small-8k-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 30

## Summary

- They share 30 benchmarks with published results for both. DeepSeek-V3 scores higher in 8 categories and Phi 3 Small 8k Instruct in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3 leads 57.6 to 31.3.
- The biggest single-benchmark swing is LiveBench Math: 73.5% for DeepSeek-V3 and 17.6% for Phi 3 Small 8k Instruct.

## Snapshot

| | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| Provider | DeepSeek | Microsoft |
| Noometry Index | 39.2 | 28.8 |
| Rank | 157 | 310 |
| Context | 164K | — |
| Input $/M | $0.24 | — |
| Output $/M | $0.90 | — |
| Weights | Open | Open |

## Coding

- DeepSeek-V3: 41.9 (#99)
- Phi 3 Small 8k Instruct: 27.4 (#313)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LiveBench Coding | 70.9% | 20.3% |
| LMArena Coding | 1368 | 1102 |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| WeirdML | 36.1% | — |
| BigCodeBench Instruct | 50% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |

## Agentic & Tool Use

- DeepSeek-V3: —
- Phi 3 Small 8k Instruct: —

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| METR Time Horizons | 49.6% | — |

## Reasoning

- DeepSeek-V3: 20.0 (#226)
- Phi 3 Small 8k Instruct: 14.2 (#319)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LiveBench Reasoning | 65.8% | 15.9% |
| LMArena Hard Prompts | 1365 | 1100 |
| LiveBench Data Analysis | 60.9% | 30.3% |
| BIG-Bench Hard | 87.5% | 79.1% |
| HellaSwag | 88.9% | 77% |
| LiveBench | 66.9% | 24% |
| WinoGrande | 85.2% | 81.5% |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| DTBench | 64.8% | — |
| LMCA | 15.5% | — |
| Adversarial NLI | — | 58.1% |
| Epoch Capabilities Index | 135.95 | — |
| ForecastBench | 59.1 | — |
| PIQA | 84.7% | — |

## Math

- DeepSeek-V3: 30.4 (#208)
- Phi 3 Small 8k Instruct: 25.4 (#247)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LiveBench Math | 73.5% | 17.6% |
| LMArena Math | 1373 | 1151 |
| OTIS Mock AIME 2024-2025 | 37.8% | — |
| Omni-MATH | 40.3% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |

## Knowledge

- DeepSeek-V3: 37.5 (#147)
- Phi 3 Small 8k Instruct: 28.5 (#237)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Expert | 1352 | 1067 |
| ARC (AI2) Challenge | 95.3% | 90.7% |
| MMLU | 87.2% | 75.7% |
| TriviaQA | 82.9% | 58.1% |
| GPQA Diamond | 67.6% | — |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| OpenBookQA | — | 88% |

## Multilingual

- DeepSeek-V3: 48.8 (#138)
- Phi 3 Small 8k Instruct: 28.6 (#267)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Non-English | 1358 | 1058 |
| LMArena Chinese | 1390 | 1061 |
| LMArena French | 1386 | 1134 |
| LMArena German | 1374 | 1080 |
| LMArena Japanese | 1333 | 966 |
| LMArena Korean | 1321 | 894 |
| LMArena Russian | 1372 | 1111 |
| LMArena Spanish | 1358 | 1111 |

## Instruction Following

- DeepSeek-V3: 73.0 (#125)
- Phi 3 Small 8k Instruct: 52.2 (#287)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LiveBench Instruction Following | 81.5% | 47.2% |
| LMArena Instruction Following | 1345 | 1087 |
| IFEval | 83.2% | — |

## Long Context

- DeepSeek-V3: 34.0 (#248)
- Phi 3 Small 8k Instruct: 33.0 (#262)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Longer Query | 1353 | 1088 |
| Fiction.LiveBench | 50% | — |

## Writing & Preference

- DeepSeek-V3: 57.6 (#124)
- Phi 3 Small 8k Instruct: 31.3 (#279)

| Benchmark | DeepSeek-V3 | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Text | 1375 | 1110 |
| LMArena Creative Writing | 1364 | 1083 |
| LMArena Multi-Turn | 1389 | 1068 |
| LiveBench Language | 49.1% | 12.9% |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |

## FAQ

### Is DeepSeek-V3 better than Phi 3 Small 8k Instruct?

DeepSeek-V3 is the stronger model overall, scoring 39.2 to 28.8 on the Noometry Index.

### Is DeepSeek-V3 or Phi 3 Small 8k Instruct better for coding?

DeepSeek-V3 scores higher on coding benchmarks: 41.9 versus 27.4 in the Noometry coding category.

### How many benchmarks do DeepSeek-V3 and Phi 3 Small 8k Instruct share?

30 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Phi 3 Small 8k Instruct has 32.