Model comparison
Gemini 3.1 Pro Preview vs GPT-4
Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 29.1 on the Noometry Index.
Last verified . 28 shared benchmarks.
GoogleSummary
- They share 28 benchmarks with published results for both. Gemini 3.1 Pro Preview scores higher in 8 categories and GPT-4 in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Gemini 3.1 Pro Preview leads 71.7 to 17.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 95.6% for Gemini 3.1 Pro Preview and 1.1% for GPT-4.
- Gemini 3.1 Pro Preview is cheaper at $2 / $12 per million input/output tokens, against $30 / $60 for GPT-4.
- Gemini 3.1 Pro Preview accepts more context: 1.05M tokens versus 8K.
Side by side
| Gemini 3.1 Pro Preview | GPT-4 | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 56.7 | 29.1 |
| Released | 2026-02-19 | 2023-03-14 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 8K |
| Max output | 66K | 8K |
| Input $ / M tokens | $2 | $30 |
| Output $ / M tokens | $12 | $60 |
| Results tracked | 71 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 42.5 (#99), GPT-4: 31.6 (#283)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| WeirdML | 72.1% | 12.4% |
| LMArena Coding | 1484 | 1254 |
| SWE-bench Verified | 75.6% | — |
| DeepSWE | 11.7% | — |
| LMArena WebDev | 1447 | — |
| SciCode | 58.9% | — |
| GSO | 22.6% | — |
| BigCodeBench Instruct | — | 46% |
| MirrorCode | 8.9% | — |
| BigCodeBench Complete | — | 57.2% |
| ALE-Bench | 1,161 | — |
| AlgoTune | 2.02 | — |
| HumanEval+ | — | 79.3% |
Agentic & Tool Use Not comparable
Gemini 3.1 Pro Preview: 37.7 (#34), GPT-4: —
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| METR Time Horizons | 77% | 36.1% |
| Terminal-Bench | 80.2% | — |
| APEX-Agents | 35.3% | — |
| τ²-bench Banking | 26% | — |
| DeepResearch Bench | 47.8% | — |
| PostTrainBench | 22% | — |
| BALROG | 57% | — |
| ExploitBench | 26.1% | — |
| GBAEval | 0.8% | — |
| GDP.pdf | 17% | — |
| LMArena Search | 1211 | — |
| Vending-Bench 2 | 3,774 | — |
Reasoning Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 71.7 (#12), GPT-4: 17.8 (#289)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| Chess Puzzles | 55% | 4% |
| LMArena Hard Prompts | 1485 | 1241 |
| Mystery Game Puzzles | 34% | 12% |
| DTBench | 97.1% | 62.7% |
| LMCA | 53.8% | 17.1% |
| Epoch Capabilities Index | 154.77 | 125.89 |
| ForecastBench | 59 | 57.8 |
| ARC-AGI-2 | 77.1% | — |
| SimpleBench | 79.6% | — |
| NYT Connections (extended) | 97.4% | — |
| ARC-AGI-1 | 98% | — |
| CritPt | 17.7% | — |
| EnigmaEval | 36.8% | — |
| Thematic Generalization | 79.4% | — |
| EBR-Bench | 14.3% | — |
| BIG-Bench Hard | — | 75.1% |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |
Math Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 62.1 (#34), GPT-4: 10.8 (#309)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 95.6% | 1.1% |
| LMArena Math | 1485 | 1269 |
| FrontierMath (Tiers 1-3) | 59.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 86.5% | — |
| ProofBench | 26% | — |
| MATH Level 5 | — | 23% |
| FrontierMath (Feb 2025 set) | 36.9% | — |
| FrontierMath Tier 4 (v1) | 16.7% | — |
| GSM8K | — | 92% |
Knowledge Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 71.8 (#3), GPT-4: 18.4 (#282)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| GPQA Diamond | 94.4% | 35.7% |
| LMArena Expert | 1485 | 1211 |
| Humanity's Last Exam | 46.4% | — |
| SimpleQA Verified | 73.5% | — |
| Vectara Hallucination Rate | 10.4% | — |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |
Multimodal Not comparable
Gemini 3.1 Pro Preview: 37.9 (#69), GPT-4: —
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| LMArena Vision | 1296 | — |
| Blueprint-Bench 2 | 26.5% | — |
| Furniture Assembly | 26.7% | — |
| LMArena Document | 1444 | — |
Multilingual Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 57.0 (#12), GPT-4: 40.6 (#215)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| LMArena Non-English | 1477 | 1246 |
| LMArena Chinese | 1529 | 1242 |
| LMArena French | 1487 | 1283 |
| LMArena German | 1491 | 1251 |
| LMArena Japanese | 1493 | 1209 |
| LMArena Korean | 1455 | 1184 |
| LMArena Russian | 1498 | 1251 |
| LMArena Spanish | 1479 | 1261 |
Instruction Following Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 77.0 (#32), GPT-4: 65.3 (#222)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1466 | 1241 |
Long Context Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 47.4 (#18), GPT-4: 37.7 (#212)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| LMArena Longer Query | 1483 | 1244 |
| CL-bench | 20.8% | — |
| CL-bench Life | 16.9% | — |
Writing & Preference Gemini 3.1 Pro Preview leads
Gemini 3.1 Pro Preview: 66.1 (#37), GPT-4: 34.9 (#268)
| Benchmark | Gemini 3.1 Pro Preview | GPT-4 |
|---|---|---|
| LMArena Text | 1481 | 1263 |
| LMArena Creative Writing | 1482 | 1244 |
| EQ-Bench Creative Writing | 1491 | 752 |
| LMArena Multi-Turn | 1488 | 1257 |
| EQ-Bench 4 | 1142 | — |
Frequently asked questions
Is Gemini 3.1 Pro Preview better than GPT-4?
Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 29.1 on the Noometry Index.
Which is cheaper, Gemini 3.1 Pro Preview or GPT-4?
Gemini 3.1 Pro Preview is cheaper. It lists at $2 per million input tokens and $12 per million output tokens; GPT-4 lists at $30 and $60.
Is Gemini 3.1 Pro Preview or GPT-4 better for coding?
Gemini 3.1 Pro Preview scores higher on coding benchmarks: 42.5 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
Gemini 3.1 Pro Preview does, with 1.05M tokens against 8K.
How many benchmarks do Gemini 3.1 Pro Preview and GPT-4 share?
28 benchmarks have published results for both models. Gemini 3.1 Pro Preview has 71 scored results on Noometry and GPT-4 has 38.