Model comparison
GPT-4 vs Llama 2-7B
GPT-4 and Llama 2-7B score almost the same on the Noometry Index (29.1 vs 29.1), so choose on price, context window or the category you care about most.
Last verified . 23 shared benchmarks.
MetaSummary
- They share 23 benchmarks with published results for both. GPT-4 scores higher in 6 categories and Llama 2-7B in 2 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Llama 2-7B leads 30.7 to 10.8.
- Llama 2-7B has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Llama 2-7B | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 29.1 | 29.1 |
| Released | 2023-03-14 | 2023-07-18 |
| Weights | Proprietary | Open |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 29 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4 leads
GPT-4: 31.6 (#283), Llama 2-7B: 29.2 (#307)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| LMArena Coding | 1254 | 1002 |
| WeirdML | 12.4% | — |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Llama 2-7B: —
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning GPT-4 leads
GPT-4: 17.8 (#289), Llama 2-7B: 15.7 (#312)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| Chess Puzzles | 4% | 0% |
| LMArena Hard Prompts | 1241 | 1009 |
| BIG-Bench Hard | 75.1% | 39.2% |
| Epoch Capabilities Index | 125.89 | 99.06 |
| HellaSwag | 95.3% | 77.2% |
| WinoGrande | 87.5% | 69.2% |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| ForecastBench | 57.8 | — |
| LAMBADA | — | 73.3% |
| PIQA | — | 78.8% |
Math Llama 2-7B leads
GPT-4: 10.8 (#309), Llama 2-7B: 30.7 (#233)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| LMArena Math | 1269 | 1042 |
| GSM8K | 92% | 16.7% |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| MATH Level 5 | 23% | — |
Knowledge Llama 2-7B leads
GPT-4: 18.4 (#282), Llama 2-7B: 28.2 (#248)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| LMArena Expert | 1211 | 1036 |
| MMLU | 86.4% | 45.8% |
| TriviaQA | 84.8% | 73.7% |
| GPQA Diamond | 35.7% | — |
| ARC (AI2) Challenge | — | 45.9% |
| BoolQ | — | 77.9% |
| OpenBookQA | — | 58.6% |
Multimodal Not comparable
GPT-4: —, Llama 2-7B: —
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| ScienceQA | — | 43.1% |
Multilingual GPT-4 leads
GPT-4: 40.6 (#215), Llama 2-7B: 23.8 (#293)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| LMArena Non-English | 1246 | 973 |
| LMArena Chinese | 1242 | 973 |
| LMArena French | 1283 | 970 |
| LMArena German | 1251 | 978 |
| LMArena Russian | 1251 | 995 |
| LMArena Spanish | 1261 | 1007 |
| LMArena Japanese | 1209 | — |
| LMArena Korean | 1184 | — |
Instruction Following GPT-4 leads
GPT-4: 65.3 (#222), Llama 2-7B: 50.8 (#298)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| LMArena Instruction Following | 1241 | 1006 |
Long Context GPT-4 leads
GPT-4: 37.7 (#212), Llama 2-7B: 30.4 (#287)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| LMArena Longer Query | 1244 | 999 |
Writing & Preference GPT-4 leads
GPT-4: 34.9 (#268), Llama 2-7B: 28.0 (#298)
| Benchmark | GPT-4 | Llama 2-7B |
|---|---|---|
| LMArena Text | 1263 | 1053 |
| LMArena Creative Writing | 1244 | 1033 |
| LMArena Multi-Turn | 1257 | 1029 |
| EQ-Bench Creative Writing | 752 | — |
Frequently asked questions
Is GPT-4 better than Llama 2-7B?
GPT-4 and Llama 2-7B score almost the same on the Noometry Index (29.1 vs 29.1), so choose on price, context window or the category you care about most.
Is GPT-4 or Llama 2-7B better for coding?
GPT-4 scores higher on coding benchmarks: 31.6 versus 29.2 in the Noometry coding category.
How many benchmarks do GPT-4 and Llama 2-7B share?
23 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Llama 2-7B has 29.