Model comparison
Step 3 vs Step 5 Preview
Step 5 Preview is the stronger model overall, scoring 47.9 to 40.5 on the Noometry Index.
Last verified . 14 shared benchmarks.
StepFunSummary
- They share 14 benchmarks with published results for both. Step 3 scores higher in 0 categories and Step 5 Preview in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in coding, where Step 5 Preview leads 51.8 to 40.1.
- Step 3 has downloadable open weights; the other is API-only.
Side by side
| Step 3 | Step 5 Preview | |
|---|---|---|
| Provider | StepFun | StepFun |
| Noometry Index | 40.5 | 47.9 |
| Released | — | 2026-09-16 |
| Weights | Open | Proprietary |
| Context window | — | 1.02M |
| Max output | — | 66K |
| Input $ / M tokens | — | $1 |
| Output $ / M tokens | — | $2.70 |
| Results tracked | 17 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 5 Preview leads
Step 3: 40.1 (#147), Step 5 Preview: 51.8 (#36)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Coding | 1367 | 1480 |
| LMArena WebDev | — | 1564 |
| SciCode | — | 58.9% |
Reasoning Step 5 Preview leads
Step 3: 28.4 (#105), Step 5 Preview: 40.0 (#57)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Hard Prompts | 1355 | 1465 |
| Kagi LLM Benchmark | 62.3% | — |
| CritPt | — | 20.9% |
Math Step 5 Preview leads
Step 3: 37.6 (#148), Step 5 Preview: 46.1 (#72)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Math | 1366 | 1470 |
| ProofBench | — | 42% |
Knowledge Step 5 Preview leads
Step 3: 36.8 (#164), Step 5 Preview: 41.2 (#112)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1333 | 1470 |
Multimodal Step 5 Preview leads
Step 3: 35.5 (#86), Step 5 Preview: 41.0 (#41)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Vision | 1177 | 1267 |
Multilingual Step 5 Preview leads
Step 3: 46.3 (#159), Step 5 Preview: 53.8 (#54)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1327 | 1432 |
| LMArena Chinese | 1397 | 1519 |
| LMArena Russian | 1331 | 1436 |
| LMArena Spanish | 1371 | 1439 |
| LMArena German | 1371 | — |
| LMArena Korean | 1269 | — |
Instruction Following Step 5 Preview leads
Step 3: 70.4 (#164), Step 5 Preview: 76.0 (#50)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1332 | 1444 |
Long Context Step 5 Preview leads
Step 3: 40.3 (#157), Step 5 Preview: 44.5 (#64)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1326 | 1455 |
Writing & Preference Step 5 Preview leads
Step 3: 54.3 (#151), Step 5 Preview: 62.8 (#71)
| Benchmark | Step 3 | Step 5 Preview |
|---|---|---|
| LMArena Text | 1350 | 1442 |
| LMArena Creative Writing | 1321 | 1410 |
| LMArena Multi-Turn | 1341 | 1447 |
Frequently asked questions
Is Step 3 better than Step 5 Preview?
Step 5 Preview is the stronger model overall, scoring 47.9 to 40.5 on the Noometry Index.
Is Step 3 or Step 5 Preview better for coding?
Step 5 Preview scores higher on coding benchmarks: 51.8 versus 40.1 in the Noometry coding category.
How many benchmarks do Step 3 and Step 5 Preview share?
14 benchmarks have published results for both models. Step 3 has 17 scored results on Noometry and Step 5 Preview has 18.