Model comparison
Llama 3.2 3B vs Qwen2.5-Coder-32B
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 28.9 on the Noometry Index. Llama 3.2 3B costs 6.2× less per token, which makes it the better buy when Qwen2.5-Coder-32B's lead doesn't matter for your workload.
Last verified . 14 shared benchmarks.
Meta
Alibaba (Qwen)Summary
- They share 14 benchmarks with published results for both. Llama 3.2 3B scores higher in 1 category and Qwen2.5-Coder-32B in 7 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Qwen2.5-Coder-32B leads 41.6 to 24.7.
- The biggest single-benchmark swing is BigCodeBench Complete: 28.3% for Llama 3.2 3B and 58% for Qwen2.5-Coder-32B.
- Llama 3.2 3B is cheaper at $0.05 / $0.33 per million input/output tokens, against $0.66 / $1 for Qwen2.5-Coder-32B.
- Llama 3.2 3B accepts more context: 131K tokens versus 33K.
Side by side
| Llama 3.2 3B | Qwen2.5-Coder-32B | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 28.9 | 33.4 |
| Released | 2024-09-24 | 2024-09-18 |
| Weights | Open | Open |
| Context window | 131K | 33K |
| Max output | 118K | 29K |
| Input $ / M tokens | $0.05 | $0.66 |
| Output $ / M tokens | $0.33 | $1 |
| Results tracked | 18 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 3.2 3B leads
Llama 3.2 3B: 27.6 (#319), Qwen2.5-Coder-32B: 22.6 (#333)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| BigCodeBench Instruct | 23.4% | 49% |
| LMArena Coding | 1098 | 1276 |
| BigCodeBench Complete | 28.3% | 58% |
| SWE-bench Verified (bash only) | — | 9% |
| Aider Polyglot | — | 16.4% |
| LiveBench Coding | — | 56.9% |
| HumanEval+ | — | 87.2% |
| MBPP+ | — | 77% |
Agentic & Tool Use Not comparable
Llama 3.2 3B: 20.1 (#143), Qwen2.5-Coder-32B: —
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 21.9% | — |
| BALROG | 10.1% | — |
Reasoning Too close to call
Llama 3.2 3B: 21.0 (#228), Qwen2.5-Coder-32B: 21.2 (#225)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Hard Prompts | 1095 | 1251 |
| LiveBench Reasoning | — | 42.1% |
| LiveBench Data Analysis | — | 49.9% |
| Epoch Capabilities Index | — | 119.49 |
| HellaSwag | — | 83% |
| LiveBench | — | 46.2% |
| WinoGrande | — | 80.8% |
Math Too close to call
Llama 3.2 3B: 32.4 (#214), Qwen2.5-Coder-32B: 33.3 (#204)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Math | 1126 | 1251 |
| LiveBench Math | — | 46.6% |
| GSM8K | — | 93% |
Knowledge Qwen2.5-Coder-32B leads
Llama 3.2 3B: 29.7 (#235), Qwen2.5-Coder-32B: 33.4 (#203)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Expert | 1090 | 1221 |
| ARC (AI2) Challenge | — | 70.5% |
| MMLU | — | 79.1% |
Multilingual Qwen2.5-Coder-32B leads
Llama 3.2 3B: 26.2 (#281), Qwen2.5-Coder-32B: 37.8 (#235)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Non-English | 1019 | 1205 |
| LMArena Chinese | 1017 | 1222 |
| LMArena Russian | 949 | 1228 |
| LMArena German | 1056 | — |
Instruction Following Qwen2.5-Coder-32B leads
Llama 3.2 3B: 56.0 (#275), Qwen2.5-Coder-32B: 61.4 (#245)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Instruction Following | 1089 | 1223 |
| LiveBench Instruction Following | — | 58.7% |
Long Context Qwen2.5-Coder-32B leads
Llama 3.2 3B: 33.4 (#261), Qwen2.5-Coder-32B: 38.0 (#208)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Longer Query | 1100 | 1251 |
Writing & Preference Qwen2.5-Coder-32B leads
Llama 3.2 3B: 24.7 (#307), Qwen2.5-Coder-32B: 41.6 (#240)
| Benchmark | Llama 3.2 3B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Text | 1110 | 1230 |
| LMArena Creative Writing | 1094 | 1174 |
| LMArena Multi-Turn | 1105 | 1222 |
| EQ-Bench Creative Writing | 595 | — |
| LiveBench Language | — | 23.3% |
Frequently asked questions
Is Llama 3.2 3B better than Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 28.9 on the Noometry Index. Llama 3.2 3B costs 6.2× less per token, which makes it the better buy when Qwen2.5-Coder-32B's lead doesn't matter for your workload.
Which is cheaper, Llama 3.2 3B or Qwen2.5-Coder-32B?
Llama 3.2 3B is cheaper. It lists at $0.05 per million input tokens and $0.33 per million output tokens; Qwen2.5-Coder-32B lists at $0.66 and $1.
Is Llama 3.2 3B or Qwen2.5-Coder-32B better for coding?
Llama 3.2 3B scores higher on coding benchmarks: 27.6 versus 22.6 in the Noometry coding category.
Which has the bigger context window?
Llama 3.2 3B does, with 131K tokens against 33K.
How many benchmarks do Llama 3.2 3B and Qwen2.5-Coder-32B share?
14 benchmarks have published results for both models. Llama 3.2 3B has 18 scored results on Noometry and Qwen2.5-Coder-32B has 31.