Model comparison
Qwen-14B vs Step 3
Step 3 is the stronger model overall, scoring 40.5 to 31.4 on the Noometry Index.
Last verified . 10 shared benchmarks.
Alibaba (Qwen)
StepFunSummary
- They share 10 benchmarks with published results for both. Qwen-14B scores higher in 0 categories and Step 3 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Step 3 leads 54.3 to 27.6.
Side by side
| Qwen-14B | Step 3 | |
|---|---|---|
| Provider | Alibaba (Qwen) | StepFun |
| Noometry Index | 31.4 | 40.5 |
| Released | 2023-09-24 | — |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 18 | 17 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 3 leads
Qwen-14B: 31.2 (#288), Step 3: 40.1 (#147)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Coding | 1071 | 1367 |
Reasoning Step 3 leads
Qwen-14B: 19.6 (#257), Step 3: 28.4 (#105)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Hard Prompts | 1027 | 1355 |
| Kagi LLM Benchmark | — | 62.3% |
| BIG-Bench Hard | 55% | — |
| Epoch Capabilities Index | 113.03 | — |
| LAMBADA | 71.1% | — |
| PIQA | 79.9% | — |
Math Step 3 leads
Qwen-14B: 31.2 (#227), Step 3: 37.6 (#148)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Math | 1068 | 1366 |
| GSM8K | 61.3% | — |
Knowledge Not comparable
Qwen-14B: —, Step 3: 36.8 (#164)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Expert | — | 1333 |
| ARC (AI2) Challenge | 84.4% | — |
| BoolQ | 86.2% | — |
| MMLU | 66.3% | — |
Multimodal Not comparable
Qwen-14B: —, Step 3: 35.5 (#86)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Vision | — | 1177 |
Multilingual Step 3 leads
Qwen-14B: 27.5 (#275), Step 3: 46.3 (#159)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Non-English | 1041 | 1327 |
| LMArena Chinese | 1077 | 1397 |
| LMArena German | — | 1371 |
| LMArena Korean | — | 1269 |
| LMArena Russian | — | 1331 |
| LMArena Spanish | — | 1371 |
Instruction Following Step 3 leads
Qwen-14B: 52.4 (#289), Step 3: 70.4 (#164)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1031 | 1332 |
Long Context Step 3 leads
Qwen-14B: 31.3 (#280), Step 3: 40.3 (#157)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Longer Query | 1028 | 1326 |
Writing & Preference Step 3 leads
Qwen-14B: 27.6 (#299), Step 3: 54.3 (#151)
| Benchmark | Qwen-14B | Step 3 |
|---|---|---|
| LMArena Text | 1051 | 1350 |
| LMArena Creative Writing | 1028 | 1321 |
| LMArena Multi-Turn | 1022 | 1341 |
Frequently asked questions
Is Qwen-14B better than Step 3?
Step 3 is the stronger model overall, scoring 40.5 to 31.4 on the Noometry Index.
Is Qwen-14B or Step 3 better for coding?
Step 3 scores higher on coding benchmarks: 40.1 versus 31.2 in the Noometry coding category.
How many benchmarks do Qwen-14B and Step 3 share?
10 benchmarks have published results for both models. Qwen-14B has 18 scored results on Noometry and Step 3 has 17.