Model comparison
Llama 2-70B vs Llama 3.1-8B
Llama 2-70B is the stronger model overall, scoring 24.1 to 22.8 on the Noometry Index.
Last verified . 26 shared benchmarks.
MetaSummary
- They share 26 benchmarks with published results for both. Llama 2-70B scores higher in 2 categories and Llama 3.1-8B in 6 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in coding, where Llama 2-70B leads 30.8 to 19.9.
- The biggest single-benchmark swing is MATH Level 5: 3.3% for Llama 2-70B and 22.9% for Llama 3.1-8B.
Side by side
| Llama 2-70B | Llama 3.1-8B | |
|---|---|---|
| Provider | Meta | Meta |
| Noometry Index | 24.1 | 22.8 |
| Released | 2023-07-18 | 2024-07-23 |
| Weights | Open | Open |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.05 |
| Output $ / M tokens | — | $0.08 |
| Results tracked | 35 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 2-70B leads
Llama 2-70B: 30.8 (#281), Llama 3.1-8B: 19.9 (#335)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| LMArena Coding | 1079 | 1195 |
| SciCode | — | 13.2% |
| WeirdML | — | 1.7% |
| BigCodeBench Instruct | — | 32.8% |
| BigCodeBench Complete | — | 40.5% |
| HumanEval+ | — | 62.8% |
| MBPP+ | — | 55.6% |
Agentic & Tool Use Not comparable
Llama 2-70B: —, Llama 3.1-8B: 22.3 (#123)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 25.8% |
| BALROG | — | 15.1% |
Reasoning Too close to call
Llama 2-70B: 13.7 (#324), Llama 3.1-8B: 14.5 (#316)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| LMArena Hard Prompts | 1073 | 1175 |
| DTBench | 41.6% | 50.9% |
| Epoch Capabilities Index | 113.59 | 116.48 |
| PIQA | 82.8% | 81.2% |
| CritPt | — | 0% |
| Chess Puzzles | — | 0% |
| LMCA | — | 5.4% |
| BIG-Bench Hard | 64.9% | — |
| CommonsenseQA 2.0 | 50% | — |
| ForecastBench | 51.4 | — |
| HellaSwag | 85.3% | — |
| LAMBADA | 78.9% | — |
| WinoGrande | 80.2% | — |
Math Llama 3.1-8B leads
Llama 2-70B: 7.5 (#320), Llama 3.1-8B: 9.5 (#311)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 0% | 1.7% |
| LMArena Math | 1091 | 1179 |
| MATH Level 5 | 3.3% | 22.9% |
| GSM8K | 69.6% | 82.4% |
| Omni-MATH | — | 13.7% |
Knowledge Too close to call
Llama 2-70B: 7.4 (#305), Llama 3.1-8B: 8.0 (#302)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| GPQA Diamond | 26.3% | 27% |
| LMArena Expert | 1039 | 1144 |
| BoolQ | 88.6% | 82.8% |
| MMLU | 69.9% | 56.1% |
| MMLU-Pro | — | 40.6% |
| GPQA (HELM) | — | 24.7% |
| ARC (AI2) Challenge | 78.3% | — |
| OpenBookQA | 60.2% | — |
| TriviaQA | 87.6% | — |
Multilingual Llama 3.1-8B leads
Llama 2-70B: 27.9 (#269), Llama 3.1-8B: 34.2 (#244)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| LMArena Non-English | 1045 | 1148 |
| LMArena Chinese | 994 | 1151 |
| LMArena French | 1089 | 1176 |
| LMArena German | 1041 | 1144 |
| LMArena Japanese | 927 | 1061 |
| LMArena Korean | 964 | 1053 |
| LMArena Russian | 1083 | 1158 |
| LMArena Spanish | 1142 | 1169 |
Instruction Following Llama 3.1-8B leads
Llama 2-70B: 55.1 (#273), Llama 3.1-8B: 59.1 (#253)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| LMArena Instruction Following | 1071 | 1159 |
| IFEval | — | 74.3% |
Long Context Llama 3.1-8B leads
Llama 2-70B: 32.2 (#265), Llama 3.1-8B: 35.8 (#233)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| LMArena Longer Query | 1062 | 1182 |
Writing & Preference Llama 2-70B leads
Llama 2-70B: 32.5 (#274), Llama 3.1-8B: 29.9 (#285)
| Benchmark | Llama 2-70B | Llama 3.1-8B |
|---|---|---|
| LMArena Text | 1115 | 1187 |
| LMArena Creative Writing | 1075 | 1154 |
| LMArena Multi-Turn | 1088 | 1172 |
| EQ-Bench Creative Writing | — | 713 |
| WildBench | — | 68.7% |
Frequently asked questions
Is Llama 2-70B better than Llama 3.1-8B?
Llama 2-70B is the stronger model overall, scoring 24.1 to 22.8 on the Noometry Index.
Is Llama 2-70B or Llama 3.1-8B better for coding?
Llama 2-70B scores higher on coding benchmarks: 30.8 versus 19.9 in the Noometry coding category.
How many benchmarks do Llama 2-70B and Llama 3.1-8B share?
26 benchmarks have published results for both models. Llama 2-70B has 35 scored results on Noometry and Llama 3.1-8B has 43.