# Gemini 3.1 Pro Preview vs GPT-4o

> Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 28.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gemini-3-1-pro-preview-vs-gpt-4o
- Last updated: 2026-10-10
- Shared benchmarks: 41

## Summary

- They share 41 benchmarks with published results for both. Gemini 3.1 Pro Preview scores higher in 10 categories and GPT-4o in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Gemini 3.1 Pro Preview leads 71.7 to 9.4.
- The biggest single-benchmark swing is ARC-AGI-1: 98% for Gemini 3.1 Pro Preview and 4.5% for GPT-4o.
- Both cost about the same: $2 input and $12 output per million tokens.
- Gemini 3.1 Pro Preview accepts more context: 1.05M tokens versus 128K.

## Snapshot

| | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| Provider | Google | OpenAI |
| Noometry Index | 56.7 | 28.5 |
| Rank | 20 | 316 |
| Context | 1.05M | 128K |
| Input $/M | $2 | $2.50 |
| Output $/M | $12 | $10 |
| Weights | Proprietary | Proprietary |

## Coding

- Gemini 3.1 Pro Preview: 42.7 (#88)
- GPT-4o: 25.1 (#323)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| SWE-bench Verified | 75.6% | 31% |
| GSO | 22.6% | 0% |
| WeirdML | 72.1% | 25.1% |
| LMArena Coding | 1483 | 1297 |
| DeepSWE | 11.7% | — |
| SWE-bench Verified (bash only) | — | 21.6% |
| Aider Polyglot | — | 45.3% |
| LMArena WebDev | 1446 | — |
| SciCode | 58.9% | — |
| BigCodeBench Instruct | — | 51.1% |
| LiveBench Coding | — | 51.4% |
| MirrorCode | 8.9% | — |
| BigCodeBench Complete | — | 61.1% |
| CadEval | — | 26% |
| ALE-Bench | 1,161 | — |
| AlgoTune | 2.02 | — |
| HumanEval+ | — | 87.2% |
| MBPP+ | — | 72.2% |

## Agentic & Tool Use

- Gemini 3.1 Pro Preview: 37.7 (#33)
- GPT-4o: 20.8 (#133)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| BALROG | 57% | 32.3% |
| LMArena Search | 1211 | 1006 |
| METR Time Horizons | 77% | 40.8% |
| Terminal-Bench | 80.2% | — |
| APEX-Agents | 35.3% | — |
| GDPval | — | 9.9% |
| TheAgentCompany | — | 8.6% |
| τ²-bench Banking | 26% | — |
| Cybench | — | 12.5% |
| DeepResearch Bench | 47.8% | — |
| PostTrainBench | 22% | — |
| ExploitBench | 26.1% | — |
| GBAEval | 0.8% | — |
| GDP.pdf | 17% | — |
| Vending-Bench 2 | 3,774 | — |

## Reasoning

- Gemini 3.1 Pro Preview: 71.7 (#9)
- GPT-4o: 9.4 (#338)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| ARC-AGI-2 | 77.1% | 0% |
| SimpleBench | 79.6% | 17.8% |
| ARC-AGI-1 | 98% | 4.5% |
| CritPt | 17.7% | 0% |
| Chess Puzzles | 55% | 13% |
| EnigmaEval | 36.8% | 0.8% |
| LMArena Hard Prompts | 1484 | 1281 |
| DTBench | 97.1% | 64.5% |
| LMCA | 53.8% | 16.6% |
| Epoch Capabilities Index | 154.92 | 128.97 |
| ForecastBench | 59 | 57.7 |
| NYT Connections (extended) | 97.4% | — |
| Thematic Generalization | 79.4% | — |
| EBR-Bench | 14.3% | — |
| LiveBench Reasoning | — | 55.8% |
| Mystery Game Puzzles | 34% | — |
| LiveBench Data Analysis | — | 60.9% |
| LiveBench | — | 55.3% |

## Math

- Gemini 3.1 Pro Preview: 61.8 (#28)
- GPT-4o: 10.2 (#303)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| FrontierMath (Tiers 1-3) | 59.6% | 0.4% |
| OTIS Mock AIME 2024-2025 | 95.6% | 6.4% |
| LMArena Math | 1485 | 1285 |
| FrontierMath (Feb 2025 set) | 36.9% | 0.3% |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 86.5% | — |
| ProofBench | 26% | — |
| Omni-MATH | — | 29.3% |
| LiveBench Math | — | 49.5% |
| MATH Level 5 | — | 53.3% |
| FrontierMath Tier 4 (v1) | 16.7% | — |

## Knowledge

- Gemini 3.1 Pro Preview: 71.8 (#3)
- GPT-4o: 28.8 (#235)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| GPQA Diamond | 94.4% | 49.2% |
| Humanity's Last Exam | 46.4% | 2.7% |
| SimpleQA Verified | 73.5% | 26% |
| Vectara Hallucination Rate | 10.4% | 9.6% |
| LMArena Expert | 1484 | 1250 |
| MMLU-Pro | — | 71.3% |
| Confabulations | — | 15.3% |
| GPQA (HELM) | — | 52% |
| MMLU | — | 88.1% |

## Multimodal

- Gemini 3.1 Pro Preview: 38.1 (#60)
- GPT-4o: 33.9 (#85)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| LMArena Vision | 1296 | 1137 |
| Video-MME | — | 71.9% |
| GeoBench | — | 71% |
| VPCT | — | 40% |
| Blueprint-Bench 2 | 26.5% | — |
| Furniture Assembly | 26.7% | — |
| LMArena Document | 1444 | — |
| ScienceQA | — | 88.5% |

## Multilingual

- Gemini 3.1 Pro Preview: 57.1 (#11)
- GPT-4o: 43.4 (#181)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| LMArena Non-English | 1475 | 1283 |
| LMArena Chinese | 1528 | 1277 |
| LMArena French | 1488 | 1303 |
| LMArena German | 1491 | 1282 |
| LMArena Japanese | 1494 | 1257 |
| LMArena Korean | 1458 | 1234 |
| LMArena Russian | 1496 | 1286 |
| LMArena Spanish | 1480 | 1292 |

## Instruction Following

- Gemini 3.1 Pro Preview: 77.1 (#31)
- GPT-4o: 66.8 (#202)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| LMArena Instruction Following | 1465 | 1278 |
| LiveBench Instruction Following | — | 68.6% |
| IFEval | — | 81.7% |

## Long Context

- Gemini 3.1 Pro Preview: 47.3 (#17)
- GPT-4o: 39.4 (#174)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| LMArena Longer Query | 1482 | 1289 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | 20.8% | — |
| CL-bench Life | 16.9% | — |

## Writing & Preference

- Gemini 3.1 Pro Preview: 66.3 (#36)
- GPT-4o: 52.8 (#160)

| Benchmark | Gemini 3.1 Pro Preview | GPT-4o |
|---|---|---|
| LMArena Text | 1480 | 1300 |
| LMArena Creative Writing | 1482 | 1291 |
| LMArena Multi-Turn | 1486 | 1302 |
| Short-Story Creative Writing | — | 81.8% |
| EQ-Bench Creative Writing | 1491 | — |
| WildBench | — | 82.8% |
| EQ-Bench 4 | 1142 | — |
| LiveBench Language | — | 47.6% |

## FAQ

### Is Gemini 3.1 Pro Preview better than GPT-4o?

Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 28.5 on the Noometry Index.

### Which is cheaper, Gemini 3.1 Pro Preview or GPT-4o?

GPT-4o is cheaper. It lists at $2.50 per million input tokens and $10 per million output tokens; Gemini 3.1 Pro Preview lists at $2 and $12.

### Is Gemini 3.1 Pro Preview or GPT-4o better for coding?

Gemini 3.1 Pro Preview scores higher on coding benchmarks: 42.7 versus 25.1 in the Noometry coding category.

### Which has the bigger context window?

Gemini 3.1 Pro Preview does, with 1.05M tokens against 128K.

### How many benchmarks do Gemini 3.1 Pro Preview and GPT-4o share?

41 benchmarks have published results for both models. Gemini 3.1 Pro Preview has 71 scored results on Noometry and GPT-4o has 72.