Model comparison
Amazon Nova Lite vs Qwen2.5-Coder-32B
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 31.9 on the Noometry Index. Amazon Nova Lite costs 7.1× less per token, which makes it the better buy when Qwen2.5-Coder-32B's lead doesn't matter for your workload.
Last verified . 20 shared benchmarks.
Alibaba (Qwen)Summary
- They share 20 benchmarks with published results for both. Amazon Nova Lite scores higher in 3 categories and Qwen2.5-Coder-32B in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in coding, where Amazon Nova Lite leads 32.5 to 22.6.
- The biggest single-benchmark swing is LiveBench Coding: 27.5% for Amazon Nova Lite and 56.9% for Qwen2.5-Coder-32B.
- Amazon Nova Lite is cheaper at $0.06 / $0.24 per million input/output tokens, against $0.66 / $1 for Qwen2.5-Coder-32B.
- Amazon Nova Lite accepts more context: 300K tokens versus 33K.
- Qwen2.5-Coder-32B has downloadable open weights; the other is API-only.
Side by side
| Amazon Nova Lite | Qwen2.5-Coder-32B | |
|---|---|---|
| Provider | Amazon | Alibaba (Qwen) |
| Noometry Index | 31.9 | 33.4 |
| Released | 2024-12-03 | 2024-09-18 |
| Weights | Proprietary | Open |
| Context window | 300K | 33K |
| Max output | 10K | 29K |
| Input $ / M tokens | $0.06 | $0.66 |
| Output $ / M tokens | $0.24 | $1 |
| Results tracked | 34 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Amazon Nova Lite leads
Amazon Nova Lite: 32.5 (#270), Qwen2.5-Coder-32B: 22.6 (#333)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LiveBench Coding | 27.5% | 56.9% |
| LMArena Coding | 1239 | 1276 |
| SWE-bench Verified (bash only) | — | 9% |
| Aider Polyglot | — | 16.4% |
| BigCodeBench Instruct | — | 49% |
| BigCodeBench Complete | — | 58% |
| ALE-Bench | 236.25 | — |
| HumanEval+ | — | 87.2% |
| MBPP+ | — | 77% |
Reasoning Qwen2.5-Coder-32B leads
Amazon Nova Lite: 19.3 (#260), Qwen2.5-Coder-32B: 21.2 (#225)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LiveBench Reasoning | 36.7% | 42.1% |
| LMArena Hard Prompts | 1220 | 1251 |
| LiveBench Data Analysis | 37.2% | 49.9% |
| LiveBench | 36.4% | 46.2% |
| Epoch Capabilities Index | — | 119.49 |
| HellaSwag | — | 83% |
| WinoGrande | — | 80.8% |
Math Qwen2.5-Coder-32B leads
Amazon Nova Lite: 27.8 (#247), Qwen2.5-Coder-32B: 33.3 (#204)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LiveBench Math | 36.7% | 46.6% |
| LMArena Math | 1227 | 1251 |
| Omni-MATH | 23.3% | — |
| GSM8K | — | 93% |
Knowledge Qwen2.5-Coder-32B leads
Amazon Nova Lite: 26.3 (#259), Qwen2.5-Coder-32B: 33.4 (#203)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Expert | 1201 | 1221 |
| MMLU | 77% | 79.1% |
| Humanity's Last Exam | 3.6% | — |
| MMLU-Pro | 60% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 39.7% | — |
| ARC (AI2) Challenge | — | 70.5% |
Multimodal Not comparable
Amazon Nova Lite: 25.5 (#123), Qwen2.5-Coder-32B: —
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Vision | 990 | — |
Multilingual Too close to call
Amazon Nova Lite: 38.0 (#232), Qwen2.5-Coder-32B: 37.8 (#235)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Non-English | 1208 | 1205 |
| LMArena Chinese | 1225 | 1222 |
| LMArena Russian | 1216 | 1228 |
| LMArena French | 1237 | — |
| LMArena German | 1229 | — |
| LMArena Japanese | 1153 | — |
| LMArena Korean | 1153 | — |
| LMArena Spanish | 1226 | — |
Instruction Following Qwen2.5-Coder-32B leads
Amazon Nova Lite: 59.3 (#255), Qwen2.5-Coder-32B: 61.4 (#245)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LiveBench Instruction Following | 54.1% | 58.7% |
| LMArena Instruction Following | 1205 | 1223 |
| IFEval | 77.6% | — |
Long Context Too close to call
Amazon Nova Lite: 37.4 (#217), Qwen2.5-Coder-32B: 38.0 (#208)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Longer Query | 1234 | 1251 |
Writing & Preference Too close to call
Amazon Nova Lite: 42.1 (#237), Qwen2.5-Coder-32B: 41.6 (#240)
| Benchmark | Amazon Nova Lite | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Text | 1229 | 1230 |
| LMArena Creative Writing | 1197 | 1174 |
| LMArena Multi-Turn | 1199 | 1222 |
| LiveBench Language | 25.9% | 23.3% |
| WildBench | 75% | — |
Frequently asked questions
Is Amazon Nova Lite better than Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 31.9 on the Noometry Index. Amazon Nova Lite costs 7.1× 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, Amazon Nova Lite or Qwen2.5-Coder-32B?
Amazon Nova Lite is cheaper. It lists at $0.06 per million input tokens and $0.24 per million output tokens; Qwen2.5-Coder-32B lists at $0.66 and $1.
Is Amazon Nova Lite or Qwen2.5-Coder-32B better for coding?
Amazon Nova Lite scores higher on coding benchmarks: 32.5 versus 22.6 in the Noometry coding category.
Which has the bigger context window?
Amazon Nova Lite does, with 300K tokens against 33K.
How many benchmarks do Amazon Nova Lite and Qwen2.5-Coder-32B share?
20 benchmarks have published results for both models. Amazon Nova Lite has 34 scored results on Noometry and Qwen2.5-Coder-32B has 31.