# Gemini 2.5 Pro vs gpt-oss-20b

> Gemini 2.5 Pro is the stronger model overall, scoring 44.9 to 32.1 on the Noometry Index. gpt-oss-20b costs 95× less per token, which makes it the better buy when Gemini 2.5 Pro's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gemini-2-5-pro-vs-gpt-oss-20b
- Last updated: 2026-10-10
- Shared benchmarks: 34

## Summary

- They share 34 benchmarks with published results for both. Gemini 2.5 Pro scores higher in 8 categories and gpt-oss-20b in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Gemini 2.5 Pro leads 63.9 to 35.8.
- The biggest single-benchmark swing is Terminal-Bench: 32.6% for Gemini 2.5 Pro and 3.4% for gpt-oss-20b.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $1.25 / $10 for Gemini 2.5 Pro.
- Gemini 2.5 Pro accepts more context: 1.05M tokens versus 131K.
- gpt-oss-20b has downloadable open weights; the other is API-only.

## Snapshot

| | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| Provider | Google | OpenAI |
| Noometry Index | 44.9 | 32.1 |
| Rank | 67 | 249 |
| Context | 1.05M | 131K |
| Input $/M | $1.25 | $0.018 |
| Output $/M | $10 | $0.09 |
| Weights | Proprietary | Open |

## Coding

- Gemini 2.5 Pro: 42.8 (#86)
- gpt-oss-20b: 37.1 (#185)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| SciCode | 42.8% | 34.4% |
| WeirdML | 54% | 40.9% |
| LMArena Coding | 1452 | 1307 |
| ALE-Bench | 785.52 | 566.05 |
| SWE-bench Verified | 57.6% | — |
| SWE-bench Verified (bash only) | 53.6% | — |
| Aider Polyglot | 83.1% | — |
| LMArena WebDev | 1227 | — |
| GSO | 3.9% | — |
| LiveBench Coding | 85.9% | — |
| CadEval | 64% | — |
| AlgoTune | 1.51 | — |

## Agentic & Tool Use

- Gemini 2.5 Pro: 29.1 (#81)
- gpt-oss-20b: 9.2 (#146)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| Terminal-Bench | 32.6% | 3.4% |
| GDPval | 23.3% | — |
| Remote Labor Index | 0.8% | — |
| TheAgentCompany | 30.3% | — |
| τ²-bench Banking | 13.7% | — |
| DeepResearch Bench | 42.8% | — |
| BALROG | 43.3% | — |
| LMArena Search | 1142 | — |
| METR Time Horizons | 55.4% | — |
| Vending-Bench 2 | 573.64 | — |

## Reasoning

- Gemini 2.5 Pro: 28.8 (#85)
- gpt-oss-20b: 18.8 (#251)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| Kagi LLM Benchmark | 70.3% | 53.2% |
| CritPt | 2% | 1.4% |
| Chess Puzzles | 20% | 4% |
| LMArena Hard Prompts | 1455 | 1274 |
| DTBench | 82.4% | 68% |
| LMCA | 34.8% | 14.5% |
| Epoch Capabilities Index | 145.32 | 137.81 |
| ARC-AGI-2 | 4.9% | — |
| SimpleBench | 62.4% | — |
| ARC-AGI-1 | 41% | — |
| EnigmaEval | 5.6% | — |
| LiveBench Reasoning | 89.8% | — |
| LiveBench Data Analysis | 79.9% | — |
| ForecastBench | 61.3 | — |
| LiveBench | 82.3% | — |

## Math

- Gemini 2.5 Pro: 31.9 (#191)
- gpt-oss-20b: 37.6 (#93)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 84.7% | 65.3% |
| Omni-MATH | 41.6% | 56.5% |
| LMArena Math | 1450 | 1317 |
| FrontierMath (Tiers 1-3) | 24.6% | — |
| FrontierMath Tier 4 | 0% | — |
| LiveBench Math | 90.2% | — |
| MATH Level 5 | 95.9% | — |
| FrontierMath (Feb 2025 set) | 14.1% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |

## Knowledge

- Gemini 2.5 Pro: 55.8 (#41)
- gpt-oss-20b: 34.6 (#186)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| GPQA Diamond | 85.3% | 60.8% |
| MMLU-Pro | 86.3% | 74% |
| GPQA (HELM) | 74.9% | 59.4% |
| LMArena Expert | 1452 | 1258 |
| Humanity's Last Exam | 21.6% | — |
| Confabulations | 10.6% | — |
| Vectara Hallucination Rate | 7% | — |

## Multimodal

- Gemini 2.5 Pro: 44.5 (#16)
- gpt-oss-20b: —

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| LMArena Vision | 1263 | — |
| GeoBench | 86% | — |
| VPCT | 48% | — |
| LMArena Document | 1421 | — |
| SpatialViz-Bench | 44.7% | — |

## Multilingual

- Gemini 2.5 Pro: 55.5 (#30)
- gpt-oss-20b: 42.4 (#192)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| LMArena Non-English | 1452 | 1269 |
| LMArena Chinese | 1507 | 1311 |
| LMArena German | 1487 | 1249 |
| LMArena Japanese | 1462 | 1245 |
| LMArena Korean | 1435 | 1239 |
| LMArena Russian | 1460 | 1279 |
| LMArena Spanish | 1473 | 1267 |
| LMArena French | 1473 | — |

## Instruction Following

- Gemini 2.5 Pro: 75.2 (#71)
- gpt-oss-20b: 62.0 (#235)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| IFEval | 84% | 73.2% |
| LMArena Instruction Following | 1436 | 1236 |
| LiveBench Instruction Following | 80.6% | — |

## Long Context

- Gemini 2.5 Pro: 59.7 (#5)
- gpt-oss-20b: 37.9 (#203)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| LMArena Longer Query | 1449 | 1251 |
| Fiction.LiveBench | 91.7% | — |

## Writing & Preference

- Gemini 2.5 Pro: 63.9 (#60)
- gpt-oss-20b: 35.8 (#260)

| Benchmark | Gemini 2.5 Pro | gpt-oss-20b |
|---|---|---|
| LMArena Text | 1458 | 1288 |
| LMArena Creative Writing | 1453 | 1201 |
| EQ-Bench Creative Writing | 1421 | 666 |
| WildBench | 85.7% | 73.7% |
| LMArena Multi-Turn | 1453 | 1268 |
| Short-Story Creative Writing | 83.8% | — |
| LiveBench Language | 67.8% | — |

## FAQ

### Is Gemini 2.5 Pro better than gpt-oss-20b?

Gemini 2.5 Pro is the stronger model overall, scoring 44.9 to 32.1 on the Noometry Index. gpt-oss-20b costs 95× less per token, which makes it the better buy when Gemini 2.5 Pro's lead doesn't matter for your workload.

### Which is cheaper, Gemini 2.5 Pro or gpt-oss-20b?

gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Gemini 2.5 Pro lists at $1.25 and $10.

### Is Gemini 2.5 Pro or gpt-oss-20b better for coding?

Gemini 2.5 Pro scores higher on coding benchmarks: 42.8 versus 37.1 in the Noometry coding category.

### Which has the bigger context window?

Gemini 2.5 Pro does, with 1.05M tokens against 131K.

### How many benchmarks do Gemini 2.5 Pro and gpt-oss-20b share?

34 benchmarks have published results for both models. Gemini 2.5 Pro has 78 scored results on Noometry and gpt-oss-20b has 34.