Model comparison
Llama 2-13B vs Llama 3-8B
Llama 2-13B is the stronger model overall, scoring 29.6 to 25.5 on the Noometry Index.
Last verified . 25 shared benchmarks.
MetaSummary
- They share 25 benchmarks with published results for both. Llama 2-13B scores higher in 2 categories and Llama 3-8B in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Llama 2-13B leads 31.1 to 8.8.
Side by side
| Llama 2-13B | Llama 3-8B | |
|---|---|---|
| Provider | Meta | Meta |
| Noometry Index | 29.6 | 25.5 |
| Released | 2023-07-18 | 2024-04-18 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 32 | 34 |
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 3-8B: 31.0 (#289)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| LMArena Coding | 1062 | 1152 |
| BigCodeBench Instruct | — | 31.9% |
| BigCodeBench Complete | — | 36.9% |
| HumanEval+ | — | 56.7% |
| MBPP+ | — | 54.8% |
Reasoning Llama 3-8B leads
Llama 2-13B: 12.8 (#337), Llama 3-8B: 14.3 (#326)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| Chess Puzzles | 0% | 0% |
| LMArena Hard Prompts | 1051 | 1133 |
| DTBench | 42.2% | 43.9% |
| Epoch Capabilities Index | 106.17 | 116.45 |
| WinoGrande | 72.8% | 75.7% |
| Adversarial NLI | — | 57.3% |
| BIG-Bench Hard | 58.2% | — |
| ForecastBench | — | 58.6 |
| HellaSwag | 80.7% | — |
| LAMBADA | 76.5% | — |
| PIQA | 80.8% | — |
Math Llama 2-13B leads
Llama 2-13B: 31.1 (#229), Llama 3-8B: 8.8 (#323)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| LMArena Math | 1065 | 1151 |
| OTIS Mock AIME 2024-2025 | — | 1.9% |
| MATH Level 5 | — | 6.1% |
| GSM8K | 36.9% | — |
Knowledge Llama 2-13B leads
Llama 2-13B: 28.1 (#249), Llama 3-8B: 7.8 (#308)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| LMArena Expert | 1030 | 1113 |
| ARC (AI2) Challenge | 60.3% | 82.8% |
| MMLU | 55.6% | 68.8% |
| OpenBookQA | 57% | 82.6% |
| TriviaQA | 79.6% | 67.7% |
| GPQA Diamond | — | 26.1% |
| BoolQ | 82.4% | — |
Multimodal Not comparable
Llama 2-13B: —, Llama 3-8B: —
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| ScienceQA | 55.8% | — |
Multilingual Llama 3-8B leads
Llama 2-13B: 26.5 (#279), Llama 3-8B: 30.8 (#261)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| LMArena Non-English | 1024 | 1098 |
| LMArena Chinese | 1001 | 1076 |
| LMArena French | 1044 | 1159 |
| LMArena German | 1009 | 1104 |
| LMArena Japanese | 894 | 967 |
| LMArena Korean | 953 | 1004 |
| LMArena Russian | 1055 | 1109 |
| LMArena Spanish | 1087 | 1173 |
Instruction Following Llama 3-8B leads
Llama 2-13B: 53.3 (#287), Llama 3-8B: 58.4 (#260)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | 1045 | 1127 |
Long Context Llama 3-8B leads
Llama 2-13B: 32.3 (#269), Llama 3-8B: 34.2 (#251)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | 1064 | 1128 |
Writing & Preference Llama 3-8B leads
Llama 2-13B: 29.8 (#289), Llama 3-8B: 37.5 (#256)
| Benchmark | Llama 2-13B | Llama 3-8B |
|---|---|---|
| LMArena Text | 1084 | 1166 |
| LMArena Creative Writing | 1047 | 1150 |
| LMArena Multi-Turn | 1050 | 1152 |
Frequently asked questions
Is Llama 2-13B better than Llama 3-8B?
Llama 2-13B is the stronger model overall, scoring 29.6 to 25.5 on the Noometry Index.
Is Llama 2-13B or Llama 3-8B better for coding?
They score almost the same on coding (30.9 vs 31.0); test both on your own repository before choosing.
How many benchmarks do Llama 2-13B and Llama 3-8B share?
25 benchmarks have published results for both models. Llama 2-13B has 32 scored results on Noometry and Llama 3-8B has 34.