Model comparison
Gemini 1.0 Pro vs GPT-4
GPT-4 is the stronger model overall, scoring 28.9 to 27.1 on the Noometry Index.
Last verified . 23 shared benchmarks.
GoogleSummary
- They share 23 benchmarks with published results for both. Gemini 1.0 Pro scores higher in 2 categories and GPT-4 in 6 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in instruction following, where GPT-4 leads 65.5 to 57.9.
- The biggest single-benchmark swing is DTBench: 45.9% for Gemini 1.0 Pro and 62.7% for GPT-4.
Side by side
| Gemini 1.0 Pro | GPT-4 | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 27.1 | 28.9 |
| Released | 2023-12-13 | 2023-03-14 |
| Weights | Proprietary | Proprietary |
| Context window | — | 8K |
| Max output | — | 8K |
| Input $ / M tokens | — | $30 |
| Output $ / M tokens | — | $60 |
| Results tracked | 24 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Gemini 1.0 Pro: 31.6 (#270), GPT-4: 31.2 (#277)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| LMArena Coding | 1108 | 1254 |
| HumanEval+ | 55.5% | 79.3% |
| WeirdML | — | 12.4% |
| BigCodeBench Instruct | — | 46% |
| BigCodeBench Complete | — | 57.2% |
| MBPP+ | 61.4% | — |
Agentic & Tool Use Not comparable
Gemini 1.0 Pro: —, GPT-4: —
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| METR Time Horizons | — | 36.1% |
Reasoning Too close to call
Gemini 1.0 Pro: 16.4 (#292), GPT-4: 17.3 (#285)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| LMArena Hard Prompts | 1109 | 1241 |
| DTBench | 45.9% | 62.7% |
| Epoch Capabilities Index | 116.97 | 125.88 |
| Chess Puzzles | — | 4% |
| Mystery Game Puzzles | — | 12% |
| LMCA | — | 17.1% |
| BIG-Bench Hard | — | 75.1% |
| ForecastBench | — | 57.8 |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |
Math GPT-4 leads
Gemini 1.0 Pro: 8.7 (#315), GPT-4: 10.1 (#304)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 1.1% |
| LMArena Math | 1132 | 1269 |
| MATH Level 5 | 11.2% | 23% |
| GSM8K | — | 92% |
Knowledge GPT-4 leads
Gemini 1.0 Pro: 15.5 (#286), GPT-4: 18.4 (#277)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| GPQA Diamond | 34% | 35.7% |
| LMArena Expert | 1059 | 1211 |
| MMLU | 70% | 86.4% |
| TriviaQA | — | 84.8% |
Multilingual GPT-4 leads
Gemini 1.0 Pro: 33.5 (#247), GPT-4: 40.8 (#210)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| LMArena Non-English | 1138 | 1246 |
| LMArena Chinese | 1124 | 1242 |
| LMArena French | 1144 | 1282 |
| LMArena German | 1125 | 1251 |
| LMArena Japanese | 1023 | 1209 |
| LMArena Russian | 1186 | 1251 |
| LMArena Spanish | 1118 | 1261 |
| LMArena Korean | — | 1184 |
Instruction Following GPT-4 leads
Gemini 1.0 Pro: 57.9 (#262), GPT-4: 65.5 (#217)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1114 | 1241 |
Long Context GPT-4 leads
Gemini 1.0 Pro: 34.3 (#244), GPT-4: 37.7 (#207)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| LMArena Longer Query | 1132 | 1244 |
Writing & Preference Gemini 1.0 Pro leads
Gemini 1.0 Pro: 36.2 (#259), GPT-4: 35.0 (#263)
| Benchmark | Gemini 1.0 Pro | GPT-4 |
|---|---|---|
| LMArena Text | 1149 | 1263 |
| LMArena Creative Writing | 1130 | 1244 |
| LMArena Multi-Turn | 1139 | 1257 |
| EQ-Bench Creative Writing | — | 752 |
Frequently asked questions
Is Gemini 1.0 Pro better than GPT-4?
GPT-4 is the stronger model overall, scoring 28.9 to 27.1 on the Noometry Index.
Is Gemini 1.0 Pro or GPT-4 better for coding?
They score almost the same on coding (31.6 vs 31.2); test both on your own repository before choosing.
How many benchmarks do Gemini 1.0 Pro and GPT-4 share?
23 benchmarks have published results for both models. Gemini 1.0 Pro has 24 scored results on Noometry and GPT-4 has 38.