Model comparison
Llama 3.2 3B vs Qwen2.5 72B Instruct
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 28.9 on the Noometry Index. Llama 3.2 3B costs 20× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.
Last verified . 16 shared benchmarks.
Meta
Alibaba (Qwen)Summary
- They share 16 benchmarks with published results for both. Llama 3.2 3B scores higher in 2 categories and Qwen2.5 72B Instruct in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Qwen2.5 72B Instruct leads 46.7 to 24.7.
- The biggest single-benchmark swing is BigCodeBench Complete: 28.3% for Llama 3.2 3B and 55.9% for Qwen2.5 72B Instruct.
- Llama 3.2 3B is cheaper at $0.05 / $0.33 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
Side by side
| Llama 3.2 3B | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 28.9 | 31.9 |
| Released | 2024-09-24 | 2024-09 |
| Weights | Open | Open |
| Context window | 131K | 131K |
| Max output | 118K | 8K |
| Input $ / M tokens | $0.05 | $1.40 |
| Output $ / M tokens | $0.33 | $5.60 |
| Results tracked | 18 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen2.5 72B Instruct leads
Llama 3.2 3B: 27.6 (#319), Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| BigCodeBench Instruct | 23.4% | 45.8% |
| LMArena Coding | 1098 | 1292 |
| BigCodeBench Complete | 28.3% | 55.9% |
| WeirdML | — | 16% |
Agentic & Tool Use Qwen2.5 72B Instruct leads
Llama 3.2 3B: 20.1 (#143), Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| BALROG | 10.1% | 16.2% |
| Berkeley Function Calling Leaderboard | 21.9% | — |
| TheAgentCompany | — | 5.7% |
| METR Time Horizons | — | 35.8% |
Reasoning Qwen2.5 72B Instruct leads
Llama 3.2 3B: 21.0 (#228), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1095 | 1271 |
| DTBench | — | 62.9% |
| LMCA | — | 13.4% |
| BIG-Bench Hard | — | 79.8% |
| Epoch Capabilities Index | — | 129 |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math Llama 3.2 3B leads
Llama 3.2 3B: 32.4 (#214), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Math | 1126 | 1283 |
| OTIS Mock AIME 2024-2025 | — | 8.1% |
| Omni-MATH | — | 33% |
| MATH Level 5 | — | 63.2% |
Knowledge Llama 3.2 3B leads
Llama 3.2 3B: 29.7 (#235), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Expert | 1090 | 1245 |
| GPQA Diamond | — | 49.1% |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| GPQA (HELM) | — | 42.6% |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual Qwen2.5 72B Instruct leads
Llama 3.2 3B: 26.2 (#281), Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1019 | 1252 |
| LMArena Chinese | 1017 | 1272 |
| LMArena German | 1056 | 1234 |
| LMArena Russian | 949 | 1264 |
| LMArena French | — | 1280 |
| LMArena Japanese | — | 1180 |
| LMArena Korean | — | 1188 |
| LMArena Spanish | — | 1256 |
Instruction Following Qwen2.5 72B Instruct leads
Llama 3.2 3B: 56.0 (#275), Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Instruction Following | 1089 | 1254 |
| IFEval | — | 80.6% |
Long Context Qwen2.5 72B Instruct leads
Llama 3.2 3B: 33.4 (#261), Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1100 | 1282 |
Writing & Preference Qwen2.5 72B Instruct leads
Llama 3.2 3B: 24.7 (#307), Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | Llama 3.2 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1110 | 1269 |
| LMArena Creative Writing | 1094 | 1221 |
| LMArena Multi-Turn | 1105 | 1272 |
| EQ-Bench Creative Writing | 595 | — |
| WildBench | — | 80.2% |
Frequently asked questions
Is Llama 3.2 3B better than Qwen2.5 72B Instruct?
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 28.9 on the Noometry Index. Llama 3.2 3B costs 20× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.
Which is cheaper, Llama 3.2 3B or Qwen2.5 72B Instruct?
Llama 3.2 3B is cheaper. It lists at $0.05 per million input tokens and $0.33 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is Llama 3.2 3B or Qwen2.5 72B Instruct better for coding?
Qwen2.5 72B Instruct scores higher on coding benchmarks: 33.2 versus 27.6 in the Noometry coding category.
Which has the bigger context window?
Both accept 131K tokens.
How many benchmarks do Llama 3.2 3B and Qwen2.5 72B Instruct share?
16 benchmarks have published results for both models. Llama 3.2 3B has 18 scored results on Noometry and Qwen2.5 72B Instruct has 43.