# DeepSeek V4 Pro vs Phi-4

> DeepSeek V4 Pro is the stronger model overall, scoring 54.2 to 30.9 on the Noometry Index. Phi-4 costs 6.2× less per token, which makes it the better buy when DeepSeek V4 Pro's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/deepseek-v4-pro-vs-phi-4
- Last updated: 2026-10-10
- Shared benchmarks: 22

## Summary

- They share 22 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and Phi-4 in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.6 to 19.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 98.6% for DeepSeek V4 Pro and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.43 / $0.87 for DeepSeek V4 Pro.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 128K.

## Snapshot

| | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| Provider | DeepSeek | Microsoft |
| Noometry Index | 54.2 | 30.9 |
| Rank | 28 | 269 |
| Context | 1M | 128K |
| Input $/M | $0.43 | $0.07 |
| Output $/M | $0.87 | $0.14 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Pro: 52.2 (#31)
- Phi-4: 33.8 (#237)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| LMArena Coding | 1470 | 1231 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| LMArena WebDev | 1582 | — |
| SciCode | 51% | — |
| WeirdML | 66.2% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 1,403 | — |

## Agentic & Tool Use

- DeepSeek V4 Pro: 32.7 (#54)
- Phi-4: 22.6 (#120)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| APEX-Agents | 47.3% | — |
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
| Vending-Bench 2 | 3,285 | — |

## Reasoning

- DeepSeek V4 Pro: 56.4 (#22)
- Phi-4: 17.0 (#287)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| Chess Puzzles | 47% | 1% |
| LMArena Hard Prompts | 1461 | 1220 |
| Epoch Capabilities Index | 155.39 | 130.43 |
| ARC-AGI-2 | 61.3% | — |
| Kagi LLM Benchmark | 53.5% | — |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 90.5% | — |
| CritPt | 18% | — |
| LiveBench Reasoning | — | 47.8% |
| Mystery Game Puzzles | 43% | — |
| DTBench | 93.9% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |
| LiveBench | — | 41.6% |

## Math

- DeepSeek V4 Pro: 64.6 (#24)
- Phi-4: 19.6 (#279)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.6% | 13.8% |
| LMArena Math | 1455 | 1246 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |

## Knowledge

- DeepSeek V4 Pro: 59.5 (#28)
- Phi-4: 32.5 (#204)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| GPQA Diamond | 91.7% | 56.1% |
| Vectara Hallucination Rate | 8.6% | 3.7% |
| LMArena Expert | 1463 | 1203 |
| SimpleQA Verified | 52.9% | — |
| Confabulations | — | 29.4% |
| MMLU | — | 84.8% |

## Multilingual

- DeepSeek V4 Pro: 54.5 (#44)
- Phi-4: 37.4 (#232)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| LMArena Non-English | 1438 | 1197 |
| LMArena Chinese | 1486 | 1211 |
| LMArena French | 1473 | 1223 |
| LMArena German | 1458 | 1222 |
| LMArena Japanese | 1448 | 1158 |
| LMArena Korean | 1452 | 1151 |
| LMArena Russian | 1452 | 1209 |
| LMArena Spanish | 1457 | 1234 |

## Instruction Following

- DeepSeek V4 Pro: 76.3 (#44)
- Phi-4: 60.6 (#246)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1448 | 1201 |
| LiveBench Instruction Following | — | 58.4% |

## Long Context

- DeepSeek V4 Pro: 45.0 (#49)
- Phi-4: 36.9 (#221)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1459 | 1217 |
| CL-bench Life | 13.5% | — |

## Writing & Preference

- DeepSeek V4 Pro: 65.6 (#45)
- Phi-4: 40.7 (#239)

| Benchmark | DeepSeek V4 Pro | Phi-4 |
|---|---|---|
| LMArena Text | 1451 | 1217 |
| LMArena Creative Writing | 1446 | 1182 |
| LMArena Multi-Turn | 1467 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench Creative Writing | 1553 | — |
| EQ-Bench 4 | 1166 | — |
| LiveBench Language | — | 25.6% |

## FAQ

### Is DeepSeek V4 Pro better than Phi-4?

DeepSeek V4 Pro is the stronger model overall, scoring 54.2 to 30.9 on the Noometry Index. Phi-4 costs 6.2× less per token, which makes it the better buy when DeepSeek V4 Pro's lead doesn't matter for your workload.

### Which is cheaper, DeepSeek V4 Pro or Phi-4?

Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; DeepSeek V4 Pro lists at $0.43 and $0.87.

### Is DeepSeek V4 Pro or Phi-4 better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.2 versus 33.8 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Pro does, with 1M tokens against 128K.

### How many benchmarks do DeepSeek V4 Pro and Phi-4 share?

22 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Phi-4 has 37.