Model comparison
Llama 2-13B vs Llama 2-70B
Llama 2-13B is the stronger model overall, scoring 29.6 to 24.4 on the Noometry Index.
Last verified . 30 shared benchmarks.
MetaSummary
- They share 30 benchmarks with published results for both. Llama 2-13B scores higher in 3 categories and Llama 2-70B in 5 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where Llama 2-13B leads 31.1 to 8.1.
Side by side
| Llama 2-13B | Llama 2-70B | |
|---|---|---|
| Provider | Meta | Meta |
| Noometry Index | 29.6 | 24.4 |
| Released | 2023-07-18 | 2023-07-18 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 32 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Llama 2-13B: 30.9 (#291), Llama 2-70B: 31.4 (#286)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Coding | 1062 | 1079 |
Reasoning Llama 2-70B leads
Llama 2-13B: 12.8 (#337), Llama 2-70B: 14.4 (#325)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Hard Prompts | 1051 | 1073 |
| DTBench | 42.2% | 41.6% |
| BIG-Bench Hard | 58.2% | 64.9% |
| Epoch Capabilities Index | 106.17 | 113.79 |
| HellaSwag | 80.7% | 85.3% |
| LAMBADA | 76.5% | 78.9% |
| PIQA | 80.8% | 82.8% |
| WinoGrande | 72.8% | 80.2% |
| Chess Puzzles | 0% | — |
| CommonsenseQA 2.0 | — | 50% |
| ForecastBench | — | 51.4 |
Math Llama 2-13B leads
Llama 2-13B: 31.1 (#229), Llama 2-70B: 8.1 (#326)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Math | 1065 | 1091 |
| GSM8K | 36.9% | 69.6% |
| OTIS Mock AIME 2024-2025 | — | 0% |
| MATH Level 5 | — | 3.3% |
Knowledge Llama 2-13B leads
Llama 2-13B: 28.1 (#249), Llama 2-70B: 7.4 (#310)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Expert | 1030 | 1039 |
| ARC (AI2) Challenge | 60.3% | 78.3% |
| BoolQ | 82.4% | 88.6% |
| MMLU | 55.6% | 69.9% |
| OpenBookQA | 57% | 60.2% |
| TriviaQA | 79.6% | 87.6% |
| GPQA Diamond | — | 26.3% |
Multimodal Not comparable
Llama 2-13B: —, Llama 2-70B: —
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| ScienceQA | 55.8% | — |
Multilingual Llama 2-70B leads
Llama 2-13B: 26.5 (#279), Llama 2-70B: 27.7 (#274)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Non-English | 1024 | 1045 |
| LMArena Chinese | 1001 | 995 |
| LMArena French | 1044 | 1090 |
| LMArena German | 1009 | 1041 |
| LMArena Japanese | 894 | 927 |
| LMArena Korean | 953 | 964 |
| LMArena Russian | 1055 | 1083 |
| LMArena Spanish | 1087 | 1143 |
Instruction Following Llama 2-70B leads
Llama 2-13B: 53.3 (#287), Llama 2-70B: 54.9 (#278)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Instruction Following | 1045 | 1071 |
Long Context Too close to call
Llama 2-13B: 32.3 (#269), Llama 2-70B: 32.3 (#270)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Longer Query | 1064 | 1062 |
Writing & Preference Llama 2-70B leads
Llama 2-13B: 29.8 (#289), Llama 2-70B: 32.3 (#279)
| Benchmark | Llama 2-13B | Llama 2-70B |
|---|---|---|
| LMArena Text | 1084 | 1115 |
| LMArena Creative Writing | 1047 | 1075 |
| LMArena Multi-Turn | 1050 | 1088 |
Frequently asked questions
Is Llama 2-13B better than Llama 2-70B?
Llama 2-13B is the stronger model overall, scoring 29.6 to 24.4 on the Noometry Index.
Is Llama 2-13B or Llama 2-70B better for coding?
They score almost the same on coding (30.9 vs 31.4); test both on your own repository before choosing.
How many benchmarks do Llama 2-13B and Llama 2-70B share?
30 benchmarks have published results for both models. Llama 2-13B has 32 scored results on Noometry and Llama 2-70B has 35.