Model comparison
Gemma 2B vs GPT-4
Gemma 2B and GPT-4 score almost the same on the Noometry Index (29.6 vs 29.1), so choose on price, context window or the category you care about most.
Last verified . 19 shared benchmarks.
GoogleSummary
- They share 19 benchmarks with published results for both. Gemma 2B scores higher in 2 categories and GPT-4 in 5 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 2B leads 30.0 to 10.8.
- Gemma 2B has downloadable open weights; the other is API-only.
Side by side
| Gemma 2B | GPT-4 | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 29.6 | 29.1 |
| Released | 2024-02-21 | 2023-03-14 |
| Weights | Open | Proprietary |
| Context window | — | 8K |
| Max output | — | 8K |
| Input $ / M tokens | — | $30 |
| Output $ / M tokens | — | $60 |
| Results tracked | 23 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4 leads
Gemma 2B: 29.4 (#305), GPT-4: 31.6 (#283)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| LMArena Coding | 1010 | 1254 |
| HumanEval+ | 20.7% | 79.3% |
| WeirdML | — | 12.4% |
| BigCodeBench Instruct | — | 46% |
| BigCodeBench Complete | — | 57.2% |
| MBPP+ | 34.1% | — |
Agentic & Tool Use Not comparable
Gemma 2B: —, GPT-4: —
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| METR Time Horizons | — | 36.1% |
Reasoning Too close to call
Gemma 2B: 18.8 (#275), GPT-4: 17.8 (#289)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| LMArena Hard Prompts | 989 | 1241 |
| BIG-Bench Hard | 35.2% | 75.1% |
| Epoch Capabilities Index | 94.2 | 125.89 |
| HellaSwag | 71.4% | 95.3% |
| WinoGrande | 65.4% | 87.5% |
| Chess Puzzles | — | 4% |
| Mystery Game Puzzles | — | 12% |
| DTBench | — | 62.7% |
| LMCA | — | 17.1% |
| ForecastBench | — | 57.8 |
| PIQA | 77.3% | — |
Math Gemma 2B leads
Gemma 2B: 30.0 (#239), GPT-4: 10.8 (#309)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| LMArena Math | 1009 | 1269 |
| GSM8K | 17.7% | 92% |
| OTIS Mock AIME 2024-2025 | — | 1.1% |
| MATH Level 5 | — | 23% |
Knowledge Not comparable
Gemma 2B: —, GPT-4: 18.4 (#282)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| MMLU | 42.3% | 86.4% |
| TriviaQA | 53.2% | 84.8% |
| GPQA Diamond | — | 35.7% |
| LMArena Expert | — | 1211 |
| ARC (AI2) Challenge | 42.1% | — |
| BoolQ | 69.4% | — |
Multilingual GPT-4 leads
Gemma 2B: 23.0 (#294), GPT-4: 40.6 (#215)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| LMArena Non-English | 958 | 1246 |
| LMArena Chinese | 986 | 1242 |
| LMArena Russian | 937 | 1251 |
| LMArena French | — | 1283 |
| LMArena German | — | 1251 |
| LMArena Japanese | — | 1209 |
| LMArena Korean | — | 1184 |
| LMArena Spanish | — | 1261 |
Instruction Following GPT-4 leads
Gemma 2B: 48.5 (#302), GPT-4: 65.3 (#222)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 970 | 1241 |
Long Context GPT-4 leads
Gemma 2B: 29.9 (#291), GPT-4: 37.7 (#212)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| LMArena Longer Query | 981 | 1244 |
Writing & Preference GPT-4 leads
Gemma 2B: 24.0 (#308), GPT-4: 34.9 (#268)
| Benchmark | Gemma 2B | GPT-4 |
|---|---|---|
| LMArena Text | 1002 | 1263 |
| LMArena Creative Writing | 987 | 1244 |
| LMArena Multi-Turn | 945 | 1257 |
| EQ-Bench Creative Writing | — | 752 |
Frequently asked questions
Is Gemma 2B better than GPT-4?
Gemma 2B and GPT-4 score almost the same on the Noometry Index (29.6 vs 29.1), so choose on price, context window or the category you care about most.
Is Gemma 2B or GPT-4 better for coding?
GPT-4 scores higher on coding benchmarks: 31.6 versus 29.4 in the Noometry coding category.
How many benchmarks do Gemma 2B and GPT-4 share?
19 benchmarks have published results for both models. Gemma 2B has 23 scored results on Noometry and GPT-4 has 38.