Model comparison
Llama 2-70B vs Qwen1.5-32B
Qwen1.5-32B is the stronger model overall, scoring 30.5 to 24.4 on the Noometry Index.
Last verified . 19 shared benchmarks.
Meta
Alibaba (Qwen)Summary
- They share 19 benchmarks with published results for both. Llama 2-70B scores higher in 0 categories and Qwen1.5-32B in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen1.5-32B leads 33.0 to 8.1.
Side by side
| Llama 2-70B | Qwen1.5-32B | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 24.4 | 30.5 |
| Released | 2023-07-18 | 2024-02-04 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 35 | 21 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Llama 2-70B: 31.4 (#286), Qwen1.5-32B: 31.7 (#282)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| LMArena Coding | 1079 | 1155 |
| BigCodeBench Instruct | — | 32.3% |
| BigCodeBench Complete | — | 42% |
Reasoning Qwen1.5-32B leads
Llama 2-70B: 14.4 (#325), Qwen1.5-32B: 21.8 (#212)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| LMArena Hard Prompts | 1073 | 1130 |
| DTBench | 41.6% | — |
| BIG-Bench Hard | 64.9% | — |
| CommonsenseQA 2.0 | 50% | — |
| Epoch Capabilities Index | 113.79 | — |
| ForecastBench | 51.4 | — |
| HellaSwag | 85.3% | — |
| LAMBADA | 78.9% | — |
| PIQA | 82.8% | — |
| WinoGrande | 80.2% | — |
Math Qwen1.5-32B leads
Llama 2-70B: 8.1 (#326), Qwen1.5-32B: 33.0 (#207)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| LMArena Math | 1091 | 1155 |
| OTIS Mock AIME 2024-2025 | 0% | — |
| MATH Level 5 | 3.3% | — |
| GSM8K | 69.6% | — |
Knowledge Qwen1.5-32B leads
Llama 2-70B: 7.4 (#310), Qwen1.5-32B: 13.5 (#296)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| GPQA Diamond | 26.3% | 30.7% |
| LMArena Expert | 1039 | 1126 |
| MMLU | 69.9% | 74.4% |
| ARC (AI2) Challenge | 78.3% | — |
| BoolQ | 88.6% | — |
| OpenBookQA | 60.2% | — |
| TriviaQA | 87.6% | — |
Multilingual Qwen1.5-32B leads
Llama 2-70B: 27.7 (#274), Qwen1.5-32B: 31.4 (#259)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| LMArena Non-English | 1045 | 1106 |
| LMArena Chinese | 995 | 1177 |
| LMArena French | 1090 | 1101 |
| LMArena German | 1041 | 1058 |
| LMArena Japanese | 927 | 1027 |
| LMArena Korean | 964 | 1008 |
| LMArena Russian | 1083 | 1073 |
| LMArena Spanish | 1143 | 1089 |
Instruction Following Qwen1.5-32B leads
Llama 2-70B: 54.9 (#278), Qwen1.5-32B: 57.7 (#265)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| LMArena Instruction Following | 1071 | 1116 |
Long Context Qwen1.5-32B leads
Llama 2-70B: 32.3 (#270), Qwen1.5-32B: 34.7 (#246)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| LMArena Longer Query | 1062 | 1146 |
Writing & Preference Qwen1.5-32B leads
Llama 2-70B: 32.3 (#279), Qwen1.5-32B: 34.2 (#271)
| Benchmark | Llama 2-70B | Qwen1.5-32B |
|---|---|---|
| LMArena Text | 1115 | 1137 |
| LMArena Creative Writing | 1075 | 1083 |
| LMArena Multi-Turn | 1088 | 1140 |
Frequently asked questions
Is Llama 2-70B better than Qwen1.5-32B?
Qwen1.5-32B is the stronger model overall, scoring 30.5 to 24.4 on the Noometry Index.
Is Llama 2-70B or Qwen1.5-32B better for coding?
They score almost the same on coding (31.4 vs 31.7); test both on your own repository before choosing.
How many benchmarks do Llama 2-70B and Qwen1.5-32B share?
19 benchmarks have published results for both models. Llama 2-70B has 35 scored results on Noometry and Qwen1.5-32B has 21.