Model comparison
Llama 2-70B vs Llama 3.1-70B
Llama 3.1-70B is the stronger model overall, scoring 29.6 to 24.4 on the Noometry Index.
Last verified . 23 shared benchmarks.
MetaSummary
- They share 23 benchmarks with published results for both. Llama 2-70B scores higher in 1 category and Llama 3.1-70B in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Llama 3.1-70B leads 24.2 to 7.4.
- The biggest single-benchmark swing is MATH Level 5: 3.3% for Llama 2-70B and 36.7% for Llama 3.1-70B.
Side by side
| Llama 2-70B | Llama 3.1-70B | |
|---|---|---|
| Provider | Meta | Meta |
| Noometry Index | 24.4 | 29.6 |
| Released | 2023-07-18 | 2024-07-23 |
| Weights | Open | Open |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.40 |
| Output $ / M tokens | — | $0.40 |
| Results tracked | 35 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 2-70B leads
Llama 2-70B: 31.4 (#286), Llama 3.1-70B: 30.3 (#296)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| LMArena Coding | 1079 | 1260 |
| WeirdML | — | 9% |
| BigCodeBench Instruct | — | 46.1% |
| BigCodeBench Complete | — | 54.8% |
Agentic & Tool Use Not comparable
Llama 2-70B: —, Llama 3.1-70B: 25.1 (#112)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| TheAgentCompany | — | 6.9% |
| BALROG | — | 27.9% |
Reasoning Llama 3.1-70B leads
Llama 2-70B: 14.4 (#325), Llama 3.1-70B: 21.6 (#220)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1073 | 1241 |
| DTBench | 41.6% | 60% |
| Epoch Capabilities Index | 113.79 | 125.92 |
| LMCA | — | 14.8% |
| BIG-Bench Hard | 64.9% | — |
| CommonsenseQA 2.0 | 50% | — |
| ForecastBench | 51.4 | — |
| HellaSwag | 85.3% | — |
| LAMBADA | 78.9% | — |
| PIQA | 82.8% | — |
| WinoGrande | 80.2% | — |
Math Llama 3.1-70B leads
Llama 2-70B: 8.1 (#326), Llama 3.1-70B: 13.5 (#304)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 0% | 3.6% |
| LMArena Math | 1091 | 1252 |
| MATH Level 5 | 3.3% | 36.7% |
| Omni-MATH | — | 21% |
| GSM8K | 69.6% | — |
Knowledge Llama 3.1-70B leads
Llama 2-70B: 7.4 (#310), Llama 3.1-70B: 24.2 (#269)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| GPQA Diamond | 26.3% | 44.2% |
| LMArena Expert | 1039 | 1209 |
| MMLU | 69.9% | 80.1% |
| MMLU-Pro | — | 65.3% |
| GPQA (HELM) | — | 42.6% |
| ARC (AI2) Challenge | 78.3% | — |
| BoolQ | 88.6% | — |
| OpenBookQA | 60.2% | — |
| TriviaQA | 87.6% | — |
Multilingual Llama 3.1-70B leads
Llama 2-70B: 27.7 (#274), Llama 3.1-70B: 38.8 (#225)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1045 | 1219 |
| LMArena Chinese | 995 | 1215 |
| LMArena French | 1090 | 1261 |
| LMArena German | 1041 | 1222 |
| LMArena Japanese | 927 | 1132 |
| LMArena Korean | 964 | 1140 |
| LMArena Russian | 1083 | 1234 |
| LMArena Spanish | 1143 | 1253 |
Instruction Following Llama 3.1-70B leads
Llama 2-70B: 54.9 (#278), Llama 3.1-70B: 65.3 (#223)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| LMArena Instruction Following | 1071 | 1231 |
| IFEval | — | 82.1% |
Long Context Llama 3.1-70B leads
Llama 2-70B: 32.3 (#270), Llama 3.1-70B: 37.6 (#214)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1062 | 1241 |
Writing & Preference Llama 3.1-70B leads
Llama 2-70B: 32.3 (#279), Llama 3.1-70B: 35.4 (#267)
| Benchmark | Llama 2-70B | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1115 | 1261 |
| LMArena Creative Writing | 1075 | 1232 |
| LMArena Multi-Turn | 1088 | 1256 |
| EQ-Bench Creative Writing | — | 784 |
| WildBench | — | 75.8% |
Frequently asked questions
Is Llama 2-70B better than Llama 3.1-70B?
Llama 3.1-70B is the stronger model overall, scoring 29.6 to 24.4 on the Noometry Index.
Is Llama 2-70B or Llama 3.1-70B better for coding?
Llama 2-70B scores higher on coding benchmarks: 31.4 versus 30.3 in the Noometry coding category.
How many benchmarks do Llama 2-70B and Llama 3.1-70B share?
23 benchmarks have published results for both models. Llama 2-70B has 35 scored results on Noometry and Llama 3.1-70B has 35.