# DeepSeek V4 Flash vs gpt-oss-120b

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 36.0 on the Noometry Index. gpt-oss-120b costs 3.7× less per token, which makes it the better buy when DeepSeek V4 Flash's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-gpt-oss-120b
- Last updated: 2026-10-10
- Shared benchmarks: 31

## Summary

- They share 31 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 8 categories and gpt-oss-120b in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Flash leads 53.7 to 19.5.
- The biggest single-benchmark swing is SimpleBench: 61.1% for DeepSeek V4 Flash and 22.1% for gpt-oss-120b.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.15 / $0.60 for DeepSeek V4 Flash.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 131K.

## Snapshot

| | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 53.6 | 36.0 |
| Rank | 32 | 210 |
| Context | 1M | 131K |
| Input $/M | $0.15 | $0.037 |
| Output $/M | $0.60 | $0.17 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Flash: 47.6 (#56)
- gpt-oss-120b: 33.2 (#251)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| SciCode | 49.9% | 36% |
| WeirdML | 63% | 48.2% |
| LMArena Coding | 1457 | 1380 |
| ALE-Bench | 1,306 | 575.62 |
| FrontierCode | 18.8% | — |
| SWE-bench Verified (bash only) | — | 26% |
| Aider Polyglot | — | 41.8% |
| LMArena WebDev | 1581 | — |
| AlgoTune | — | 1.41 |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- gpt-oss-120b: 12.1 (#145)

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

## Reasoning

- DeepSeek V4 Flash: 53.7 (#25)
- gpt-oss-120b: 19.5 (#236)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| SimpleBench | 61.1% | 22.1% |
| Kagi LLM Benchmark | 52.2% | 58.6% |
| CritPt | 16.6% | 1.1% |
| Chess Puzzles | 33% | 20% |
| LMArena Hard Prompts | 1443 | 1364 |
| Mystery Game Puzzles | 34% | 2% |
| DTBench | 90.9% | 76.3% |
| LMCA | 41.7% | 22.1% |
| Epoch Capabilities Index | 154.51 | 139.94 |
| ARC-AGI-2 | 61.4% | — |
| NYT Connections (extended) | 89.6% | — |
| ARC-AGI-1 | 89% | — |
| Surface Evolver Bench | — | 25% |

## Math

- DeepSeek V4 Flash: 60.0 (#31)
- gpt-oss-120b: 50.6 (#43)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 94.4% | 88.9% |
| LMArena Math | 1427 | 1388 |
| FrontierMath (Tiers 1-3) | 57.5% | — |
| FrontierMath Tier 4 | 24.4% | — |
| MathArena Final-Answer Competitions | 76.5% | — |
| ProofBench | 56% | — |
| Omni-MATH | — | 68.8% |

## Knowledge

- DeepSeek V4 Flash: 55.5 (#43)
- gpt-oss-120b: 42.3 (#89)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| GPQA Diamond | 91% | 75.8% |
| LMArena Expert | 1441 | 1356 |
| SimpleQA Verified | 33.6% | — |
| MMLU-Pro | — | 79.5% |
| Confabulations | — | 15.7% |
| Vectara Hallucination Rate | — | 14.2% |
| GPQA (HELM) | — | 68.4% |

## Multilingual

- DeepSeek V4 Flash: 53.2 (#71)
- gpt-oss-120b: 48.2 (#142)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| LMArena Non-English | 1420 | 1350 |
| LMArena Chinese | 1467 | 1384 |
| LMArena French | 1438 | 1370 |
| LMArena German | 1418 | 1355 |
| LMArena Japanese | 1403 | 1332 |
| LMArena Korean | 1383 | 1283 |
| LMArena Russian | 1428 | 1343 |
| LMArena Spanish | 1437 | 1388 |

## Instruction Following

- DeepSeek V4 Flash: 75.1 (#76)
- gpt-oss-120b: 69.5 (#168)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| LMArena Instruction Following | 1421 | 1318 |
| IFEval | — | 83.6% |

## Long Context

- DeepSeek V4 Flash: 43.8 (#81)
- gpt-oss-120b: 31.4 (#273)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| LMArena Longer Query | 1435 | 1319 |
| Fiction.LiveBench | — | 44.4% |

## Writing & Preference

- DeepSeek V4 Flash: 64.0 (#59)
- gpt-oss-120b: 46.7 (#213)

| Benchmark | DeepSeek V4 Flash | gpt-oss-120b |
|---|---|---|
| LMArena Text | 1432 | 1366 |
| LMArena Creative Writing | 1403 | 1274 |
| EQ-Bench Creative Writing | 1559 | 961 |
| LMArena Multi-Turn | 1449 | 1340 |
| Short-Story Creative Writing | — | 77.1% |
| WildBench | — | 84.5% |

## FAQ

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

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 36.0 on the Noometry Index. gpt-oss-120b costs 3.7× less per token, which makes it the better buy when DeepSeek V4 Flash's lead doesn't matter for your workload.

### Which is cheaper, DeepSeek V4 Flash 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 Flash lists at $0.15 and $0.60.

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

DeepSeek V4 Flash scores higher on coding benchmarks: 47.6 versus 33.2 in the Noometry coding category.

### Which has the bigger context window?

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

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

31 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and gpt-oss-120b has 48.