# DeepSeek V4 Pro vs gpt-oss-120b

> DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 36.3 on the Noometry Index. gpt-oss-120b costs 14× 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-oss-120b
- Last updated: 2026-10-10
- Shared benchmarks: 34

## Summary

- They share 34 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and gpt-oss-120b in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.5 to 20.0.
- The biggest single-benchmark swing is APEX-Agents: 47.3% for DeepSeek V4 Pro and 4.4% for gpt-oss-120b.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.66 / $1.98 for DeepSeek V4 Pro.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 131K.

## Snapshot

| | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 54.3 | 36.3 |
| Rank | 31 | 217 |
| Context | 1M | 131K |
| Input $/M | $0.66 | $0.037 |
| Output $/M | $1.98 | $0.17 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Pro: 52.4 (#34)
- gpt-oss-120b: 33.5 (#256)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| SciCode | 51% | 36% |
| WeirdML | 66.2% | 48.2% |
| LMArena Coding | 1470 | 1380 |
| ALE-Bench | 1,403 | 575.62 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| SWE-bench Verified (bash only) | — | 26% |
| Aider Polyglot | — | 41.8% |
| LMArena WebDev | 1582 | — |
| AlgoTune | — | 1.41 |

## Agentic & Tool Use

- DeepSeek V4 Pro: 32.8 (#58)
- gpt-oss-120b: 12.2 (#153)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| APEX-Agents | 47.3% | 4.4% |
| Vending-Bench 2 | 3,285 | -21.53 |
| Terminal-Bench | — | 18.7% |
| METR Time Horizons | — | 56.6% |

## Reasoning

- DeepSeek V4 Pro: 56.5 (#24)
- gpt-oss-120b: 20.0 (#245)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| Kagi LLM Benchmark | 53.5% | 58.6% |
| CritPt | 18% | 1.1% |
| Chess Puzzles | 47% | 20% |
| LMArena Hard Prompts | 1461 | 1364 |
| Mystery Game Puzzles | 43% | 2% |
| DTBench | 93.9% | 76.3% |
| LMCA | 45.5% | 22.1% |
| Surface Evolver Bench | 40% | 25% |
| Epoch Capabilities Index | 155.31 | 139.93 |
| ARC-AGI-2 | 61.3% | — |
| SimpleBench | — | 22.1% |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 90.5% | — |
| ForecastBench | 56.1 | — |

## Math

- DeepSeek V4 Pro: 64.8 (#30)
- gpt-oss-120b: 52.5 (#50)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.6% | 88.9% |
| LMArena Math | 1455 | 1389 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| Omni-MATH | — | 68.8% |

## Knowledge

- DeepSeek V4 Pro: 59.5 (#31)
- gpt-oss-120b: 42.4 (#96)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| GPQA Diamond | 91.7% | 75.8% |
| Vectara Hallucination Rate | 8.6% | 14.2% |
| LMArena Expert | 1464 | 1356 |
| SimpleQA Verified | 52.9% | — |
| MMLU-Pro | — | 79.5% |
| Confabulations | — | 15.7% |
| GPQA (HELM) | — | 68.4% |

## Multilingual

- DeepSeek V4 Pro: 54.4 (#45)
- gpt-oss-120b: 48.0 (#147)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| LMArena Non-English | 1439 | 1351 |
| LMArena Chinese | 1486 | 1385 |
| LMArena French | 1472 | 1369 |
| LMArena German | 1458 | 1353 |
| LMArena Japanese | 1445 | 1331 |
| LMArena Korean | 1447 | 1282 |
| LMArena Russian | 1453 | 1343 |
| LMArena Spanish | 1458 | 1389 |

## Instruction Following

- DeepSeek V4 Pro: 76.1 (#47)
- gpt-oss-120b: 69.3 (#173)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| LMArena Instruction Following | 1448 | 1318 |
| IFEval | — | 83.6% |

## Long Context

- DeepSeek V4 Pro: 45.0 (#51)
- gpt-oss-120b: 31.4 (#278)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| LMArena Longer Query | 1458 | 1319 |
| Fiction.LiveBench | — | 44.4% |
| CL-bench Life | 13.5% | — |

## Writing & Preference

- DeepSeek V4 Pro: 65.5 (#46)
- gpt-oss-120b: 46.5 (#217)

| Benchmark | DeepSeek V4 Pro | gpt-oss-120b |
|---|---|---|
| LMArena Text | 1451 | 1365 |
| LMArena Creative Writing | 1446 | 1275 |
| EQ-Bench Creative Writing | 1553 | 961 |
| LMArena Multi-Turn | 1467 | 1340 |
| Short-Story Creative Writing | — | 77.1% |
| WildBench | — | 84.5% |
| EQ-Bench 4 | 1166 | — |

## FAQ

### Is DeepSeek V4 Pro better than gpt-oss-120b?

DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 36.3 on the Noometry Index. gpt-oss-120b costs 14× 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-oss-120b?

gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; DeepSeek V4 Pro lists at $0.66 and $1.98.

### Is DeepSeek V4 Pro or gpt-oss-120b better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 33.5 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Pro does, with 1M tokens against 131K.

### How many benchmarks do DeepSeek V4 Pro and gpt-oss-120b share?

34 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and gpt-oss-120b has 48.