# DeepSeek V4 Pro vs GPT-4o mini

> DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 25.5 on the Noometry Index. GPT-4o mini costs 3.8× 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-gpt-4o-mini
- Last updated: 2026-10-10
- Shared benchmarks: 30

## Summary

- They share 30 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and GPT-4o mini in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.8 to 10.4.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 98.6% for DeepSeek V4 Pro and 6.9% for GPT-4o mini.
- GPT-4o mini is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.66 / $1.98 for DeepSeek V4 Pro.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 128K.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 54.3 | 25.5 |
| Rank | 31 | 343 |
| Context | 1M | 128K |
| Input $/M | $0.66 | $0.15 |
| Output $/M | $1.98 | $0.60 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek V4 Pro: 52.4 (#34)
- GPT-4o mini: 22.0 (#335)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| WeirdML | 66.2% | 11.8% |
| LMArena Coding | 1470 | 1290 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| Aider Polyglot | — | 3.6% |
| LMArena WebDev | 1582 | — |
| SciCode | 51% | — |
| BigCodeBench Instruct | — | 46.1% |
| LiveBench Coding | — | 43.1% |
| BigCodeBench Complete | — | 57.4% |
| ALE-Bench | 1,403 | — |
| HumanEval+ | — | 83.5% |
| MBPP+ | — | 72.2% |

## Agentic & Tool Use

- DeepSeek V4 Pro: 32.8 (#58)
- GPT-4o mini: 27.5 (#101)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| APEX-Agents | 47.3% | — |
| BALROG | — | 17.4% |
| Vending-Bench 2 | 3,285 | — |

## Reasoning

- DeepSeek V4 Pro: 56.5 (#24)
- GPT-4o mini: 8.7 (#347)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| ARC-AGI-2 | 61.3% | 0% |
| Kagi LLM Benchmark | 53.5% | 28.8% |
| Chess Puzzles | 47% | 0% |
| LMArena Hard Prompts | 1461 | 1267 |
| Mystery Game Puzzles | 43% | 12% |
| DTBench | 93.9% | 54.4% |
| LMCA | 45.5% | 10.4% |
| Epoch Capabilities Index | 155.31 | 126.56 |
| SimpleBench | — | 10.7% |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 90.5% | — |
| CritPt | 18% | — |
| LiveBench Reasoning | — | 32.8% |
| LiveBench Data Analysis | — | 50% |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |
| LiveBench | — | 41.3% |
| PIQA | — | 88.7% |

## Math

- DeepSeek V4 Pro: 64.8 (#30)
- GPT-4o mini: 10.4 (#314)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 0.7% |
| OTIS Mock AIME 2024-2025 | 98.6% | 6.9% |
| LMArena Math | 1455 | 1267 |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| Omni-MATH | — | 28% |
| LiveBench Math | — | 36.3% |
| MATH Level 5 | — | 52.6% |
| GSM8K | — | 91.3% |

## Knowledge

- DeepSeek V4 Pro: 59.5 (#31)
- GPT-4o mini: 17.7 (#284)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| GPQA Diamond | 91.7% | 37.7% |
| SimpleQA Verified | 52.9% | 8.3% |
| LMArena Expert | 1464 | 1235 |
| MMLU-Pro | — | 60.3% |
| Confabulations | — | 37.2% |
| Vectara Hallucination Rate | 8.6% | — |
| GPQA (HELM) | — | 36.8% |
| BoolQ | — | 88.7% |
| MMLU | — | 81.8% |

## Multimodal

- DeepSeek V4 Pro: —
- GPT-4o mini: 25.9 (#122)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| LMArena Vision | — | 1066 |
| Video-MME | — | 64.8% |
| GeoBench | — | 64% |
| VPCT | — | 34% |

## Multilingual

- DeepSeek V4 Pro: 54.4 (#45)
- GPT-4o mini: 42.0 (#199)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| LMArena Non-English | 1439 | 1266 |
| LMArena Chinese | 1486 | 1265 |
| LMArena French | 1472 | 1297 |
| LMArena German | 1458 | 1272 |
| LMArena Japanese | 1445 | 1216 |
| LMArena Korean | 1447 | 1195 |
| LMArena Russian | 1453 | 1275 |
| LMArena Spanish | 1458 | 1276 |

## Instruction Following

- DeepSeek V4 Pro: 76.1 (#47)
- GPT-4o mini: 61.9 (#239)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| LMArena Instruction Following | 1448 | 1258 |
| LiveBench Instruction Following | — | 56.8% |
| IFEval | — | 78.2% |

## Long Context

- DeepSeek V4 Pro: 45.0 (#51)
- GPT-4o mini: 39.1 (#186)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| LMArena Longer Query | 1458 | 1289 |
| CL-bench Life | 13.5% | — |

## Writing & Preference

- DeepSeek V4 Pro: 65.5 (#46)
- GPT-4o mini: 39.5 (#248)

| Benchmark | DeepSeek V4 Pro | GPT-4o mini |
|---|---|---|
| LMArena Text | 1451 | 1286 |
| LMArena Creative Writing | 1446 | 1268 |
| EQ-Bench Creative Writing | 1553 | 873 |
| LMArena Multi-Turn | 1467 | 1285 |
| Short-Story Creative Writing | — | 67.2% |
| WildBench | — | 79.1% |
| EQ-Bench 4 | 1166 | — |
| LiveBench Language | — | 28.6% |

## FAQ

### Is DeepSeek V4 Pro better than GPT-4o mini?

DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 25.5 on the Noometry Index. GPT-4o mini costs 3.8× 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 GPT-4o mini?

GPT-4o mini is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; DeepSeek V4 Pro lists at $0.66 and $1.98.

### Is DeepSeek V4 Pro or GPT-4o mini better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 22.0 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 GPT-4o mini share?

30 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and GPT-4o mini has 60.