Model comparison
Ministral 3B vs Qwen2.5 72B Instruct
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 26.2 on the Noometry Index. Ministral 3B costs 24× less per token, which makes it the better buy when Qwen2.5 72B Instruct's lead doesn't matter for your workload.
Last verified . 5 shared benchmarks.
Mistral AI
Alibaba (Qwen)Summary
- They share 5 benchmarks with published results for both. Ministral 3B scores higher in 1 category and Qwen2.5 72B Instruct in 2 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen2.5 72B Instruct leads 27.0 to 10.4.
- The biggest single-benchmark swing is MATH Level 5: 14.4% for Ministral 3B and 63.2% for Qwen2.5 72B Instruct.
- Ministral 3B is cheaper at $0.10 / $0.10 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
Side by side
| Ministral 3B | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Mistral AI | Alibaba (Qwen) |
| Noometry Index | 26.2 | 31.9 |
| Released | 2024-10-01 | 2024-09 |
| Weights | Open | Open |
| Context window | 131K | 131K |
| Max output | 262K | 8K |
| Input $ / M tokens | $0.10 | $1.40 |
| Output $ / M tokens | $0.10 | $5.60 |
| Results tracked | 6 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
Ministral 3B: —, Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | — | 16% |
| BigCodeBench Instruct | — | 45.8% |
| LMArena Coding | — | 1292 |
| BigCodeBench Complete | — | 55.9% |
Agentic & Tool Use Not comparable
Ministral 3B: —, Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
Reasoning Qwen2.5 72B Instruct leads
Ministral 3B: 18.4 (#282), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| DTBench | 51.7% | 62.9% |
| LMCA | 5.5% | 13.4% |
| Epoch Capabilities Index | 118.1 | 129 |
| LMArena Hard Prompts | — | 1271 |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math Ministral 3B leads
Ministral 3B: 26.6 (#258), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| MATH Level 5 | 14.4% | 63.2% |
| OTIS Mock AIME 2024-2025 | — | 8.1% |
| Omni-MATH | — | 33% |
| LMArena Math | — | 1283 |
Knowledge Qwen2.5 72B Instruct leads
Ministral 3B: 10.4 (#302), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 25.3% | 49.1% |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| Vectara Hallucination Rate | 7.3% | — |
| GPQA (HELM) | — | 42.6% |
| LMArena Expert | — | 1245 |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual Not comparable
Ministral 3B: —, Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | Ministral 3B | 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
Ministral 3B: —, Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | — | 80.6% |
| LMArena Instruction Following | — | 1254 |
Long Context Not comparable
Ministral 3B: —, Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | — | 1282 |
Writing & Preference Not comparable
Ministral 3B: —, Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | Ministral 3B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | — | 1269 |
| LMArena Creative Writing | — | 1221 |
| WildBench | — | 80.2% |
| LMArena Multi-Turn | — | 1272 |
Frequently asked questions
Is Ministral 3B better than Qwen2.5 72B Instruct?
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 26.2 on the Noometry Index. Ministral 3B costs 24× 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, Ministral 3B or Qwen2.5 72B Instruct?
Ministral 3B is cheaper. It lists at $0.10 per million input tokens and $0.10 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Which has the bigger context window?
Both accept 131K tokens.
How many benchmarks do Ministral 3B and Qwen2.5 72B Instruct share?
5 benchmarks have published results for both models. Ministral 3B has 6 scored results on Noometry and Qwen2.5 72B Instruct has 43.