Model comparison
Mixtral 8x22B vs Qwen2.5-Coder-32B
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 27.1 on the Noometry Index.
Last verified . 18 shared benchmarks.
Mistral AI
Alibaba (Qwen)Summary
- They share 18 benchmarks with published results for both. Mixtral 8x22B scores higher in 1 category and Qwen2.5-Coder-32B in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen2.5-Coder-32B leads 33.4 to 15.1.
- The biggest single-benchmark swing is BigCodeBench Instruct: 40.6% for Mixtral 8x22B and 49% for Qwen2.5-Coder-32B.
- Qwen2.5-Coder-32B is cheaper at $0.66 / $1 per million input/output tokens, against $2 / $6 for Mixtral 8x22B.
- Mixtral 8x22B accepts more context: 64K tokens versus 33K.
Side by side
| Mixtral 8x22B | Qwen2.5-Coder-32B | |
|---|---|---|
| Provider | Mistral AI | Alibaba (Qwen) |
| Noometry Index | 27.1 | 33.4 |
| Released | 2024-04-17 | 2024-09-18 |
| Weights | Open | Open |
| Context window | 64K | 33K |
| Max output | 64K | 29K |
| Input $ / M tokens | $2 | $0.66 |
| Output $ / M tokens | $6 | $1 |
| Results tracked | 34 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mixtral 8x22B leads
Mixtral 8x22B: 24.2 (#329), Qwen2.5-Coder-32B: 22.6 (#333)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| BigCodeBench Instruct | 40.6% | 49% |
| LMArena Coding | 1166 | 1276 |
| BigCodeBench Complete | 50.2% | 58% |
| HumanEval+ | 72% | 87.2% |
| MBPP+ | 64.3% | 77% |
| SWE-bench Verified (bash only) | — | 9% |
| Aider Polyglot | — | 16.4% |
| WeirdML | 3.2% | — |
| LiveBench Coding | — | 56.9% |
Agentic & Tool Use Not comparable
Mixtral 8x22B: 23.1 (#127), Qwen2.5-Coder-32B: —
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| Cybench | 7.5% | — |
Reasoning Qwen2.5-Coder-32B leads
Mixtral 8x22B: 19.9 (#248), Qwen2.5-Coder-32B: 21.2 (#225)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Hard Prompts | 1150 | 1251 |
| Epoch Capabilities Index | 122.03 | 119.49 |
| LiveBench Reasoning | — | 42.1% |
| DTBench | 55.1% | — |
| LiveBench Data Analysis | — | 49.9% |
| ForecastBench | 56.3 | — |
| HellaSwag | — | 83% |
| LiveBench | — | 46.2% |
| WinoGrande | — | 80.8% |
Math Qwen2.5-Coder-32B leads
Mixtral 8x22B: 22.9 (#275), Qwen2.5-Coder-32B: 33.3 (#204)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Math | 1184 | 1251 |
| Omni-MATH | 16.3% | — |
| LiveBench Math | — | 46.6% |
| MATH Level 5 | 24.2% | — |
| GSM8K | — | 93% |
Knowledge Qwen2.5-Coder-32B leads
Mixtral 8x22B: 15.1 (#293), Qwen2.5-Coder-32B: 33.4 (#203)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Expert | 1113 | 1221 |
| MMLU | 77.8% | 79.1% |
| GPQA Diamond | 34.1% | — |
| MMLU-Pro | 46% | — |
| GPQA (HELM) | 33.4% | — |
| ARC (AI2) Challenge | — | 70.5% |
Multilingual Qwen2.5-Coder-32B leads
Mixtral 8x22B: 32.8 (#255), Qwen2.5-Coder-32B: 37.8 (#235)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Non-English | 1128 | 1205 |
| LMArena Chinese | 1116 | 1222 |
| LMArena Russian | 1158 | 1228 |
| LMArena French | 1166 | — |
| LMArena German | 1141 | — |
| LMArena Japanese | 1037 | — |
| LMArena Korean | 1057 | — |
| LMArena Spanish | 1151 | — |
Instruction Following Qwen2.5-Coder-32B leads
Mixtral 8x22B: 57.7 (#266), Qwen2.5-Coder-32B: 61.4 (#245)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Instruction Following | 1147 | 1223 |
| LiveBench Instruction Following | — | 58.7% |
| IFEval | 72.4% | — |
Long Context Qwen2.5-Coder-32B leads
Mixtral 8x22B: 34.7 (#247), Qwen2.5-Coder-32B: 38.0 (#208)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Longer Query | 1144 | 1251 |
Writing & Preference Qwen2.5-Coder-32B leads
Mixtral 8x22B: 36.9 (#262), Qwen2.5-Coder-32B: 41.6 (#240)
| Benchmark | Mixtral 8x22B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Text | 1162 | 1230 |
| LMArena Creative Writing | 1141 | 1174 |
| LMArena Multi-Turn | 1130 | 1222 |
| WildBench | 71.1% | — |
| LiveBench Language | — | 23.3% |
Frequently asked questions
Is Mixtral 8x22B better than Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 27.1 on the Noometry Index.
Which is cheaper, Mixtral 8x22B or Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is cheaper. It lists at $0.66 per million input tokens and $1 per million output tokens; Mixtral 8x22B lists at $2 and $6.
Is Mixtral 8x22B or Qwen2.5-Coder-32B better for coding?
Mixtral 8x22B scores higher on coding benchmarks: 24.2 versus 22.6 in the Noometry coding category.
Which has the bigger context window?
Mixtral 8x22B does, with 64K tokens against 33K.
How many benchmarks do Mixtral 8x22B and Qwen2.5-Coder-32B share?
18 benchmarks have published results for both models. Mixtral 8x22B has 34 scored results on Noometry and Qwen2.5-Coder-32B has 31.