Model comparison
Qwen2.5 32B Instruct vs Qwen2.5 72B Instruct
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 30.1 on the Noometry Index. Qwen2.5 32B Instruct costs 2.0× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.
Last verified . 6 shared benchmarks.
Alibaba (Qwen)Summary
- They share 6 benchmarks with published results for both. Qwen2.5 32B Instruct scores higher in 1 category and Qwen2.5 72B Instruct in 3 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in coding, where Qwen2.5 32B Instruct leads 38.7 to 33.2.
- The biggest single-benchmark swing is MATH Level 5: 56.1% for Qwen2.5 32B Instruct and 63.2% for Qwen2.5 72B Instruct.
- Qwen2.5 32B Instruct is cheaper at $0.70 / $2.80 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
Side by side
| Qwen2.5 32B Instruct | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Alibaba (Qwen) | Alibaba (Qwen) |
| Noometry Index | 30.1 | 31.9 |
| Released | 2024-09 | 2024-09 |
| Weights | Open | Open |
| Context window | 131K | 131K |
| Max output | 8K | 8K |
| Input $ / M tokens | $0.70 | $1.40 |
| Output $ / M tokens | $2.80 | $5.60 |
| Results tracked | 7 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen2.5 32B Instruct leads
Qwen2.5 32B Instruct: 38.7 (#169), Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| BigCodeBench Instruct | 45% | 45.8% |
| BigCodeBench Complete | 52.3% | 55.9% |
| WeirdML | — | 16% |
| LMArena Coding | — | 1292 |
Agentic & Tool Use Not comparable
Qwen2.5 32B Instruct: —, Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
Reasoning Qwen2.5 72B Instruct leads
Qwen2.5 32B Instruct: 19.2 (#266), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| Epoch Capabilities Index | 128.52 | 129 |
| Chess Puzzles | 0% | — |
| LMArena Hard Prompts | — | 1271 |
| DTBench | — | 62.9% |
| LMCA | — | 13.4% |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math Qwen2.5 72B Instruct leads
Qwen2.5 32B Instruct: 16.2 (#296), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.4% | 8.1% |
| MATH Level 5 | 56.1% | 63.2% |
| Omni-MATH | — | 33% |
| LMArena Math | — | 1283 |
Knowledge Qwen2.5 72B Instruct leads
Qwen2.5 32B Instruct: 24.9 (#266), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 46.1% | 49.1% |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| GPQA (HELM) | — | 42.6% |
| LMArena Expert | — | 1245 |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual Not comparable
Qwen2.5 32B Instruct: —, Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | — | 1252 |
| LMArena Chinese | — | 1272 |
| LMArena French | — | 1280 |
| LMArena German | — | 1234 |
| LMArena Japanese | — | 1180 |
| LMArena Korean | — | 1188 |
| LMArena Russian | — | 1264 |
| LMArena Spanish | — | 1256 |
Instruction Following Not comparable
Qwen2.5 32B Instruct: —, Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | — | 80.6% |
| LMArena Instruction Following | — | 1254 |
Long Context Not comparable
Qwen2.5 32B Instruct: —, Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | — | 1282 |
Writing & Preference Not comparable
Qwen2.5 32B Instruct: —, Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | Qwen2.5 32B Instruct | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | — | 1269 |
| LMArena Creative Writing | — | 1221 |
| WildBench | — | 80.2% |
| LMArena Multi-Turn | — | 1272 |
Frequently asked questions
Is Qwen2.5 32B Instruct better than Qwen2.5 72B Instruct?
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 30.1 on the Noometry Index. Qwen2.5 32B Instruct costs 2.0× 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, Qwen2.5 32B Instruct or Qwen2.5 72B Instruct?
Qwen2.5 32B Instruct is cheaper. It lists at $0.70 per million input tokens and $2.80 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is Qwen2.5 32B Instruct or Qwen2.5 72B Instruct better for coding?
Qwen2.5 32B Instruct scores higher on coding benchmarks: 38.7 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
Both accept 131K tokens.
How many benchmarks do Qwen2.5 32B Instruct and Qwen2.5 72B Instruct share?
6 benchmarks have published results for both models. Qwen2.5 32B Instruct has 7 scored results on Noometry and Qwen2.5 72B Instruct has 43.