Model comparison
DeepSeek-V3.1 vs Step 5 Preview
Step 5 Preview is the stronger model overall, scoring 47.9 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 3.4× less per token, which makes it the better buy when Step 5 Preview's lead doesn't matter for your workload.
Last verified . 13 shared benchmarks.
DeepSeek
StepFunSummary
- They share 13 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 1 category and Step 5 Preview in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Step 5 Preview leads 40.0 to 27.9.
- DeepSeek-V3.1 is cheaper at $0.25 / $0.95 per million input/output tokens, against $1 / $2.70 for Step 5 Preview.
- Step 5 Preview accepts more context: 1.02M tokens versus 164K.
- DeepSeek-V3.1 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.1 | Step 5 Preview | |
|---|---|---|
| Provider | DeepSeek | StepFun |
| Noometry Index | 42.8 | 47.9 |
| Released | 2025-08-21 | 2026-09-16 |
| Weights | Open | Proprietary |
| Context window | 164K | 1.02M |
| Max output | 8K | 66K |
| Input $ / M tokens | $0.25 | $1 |
| Output $ / M tokens | $0.95 | $2.70 |
| Results tracked | 27 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 5 Preview leads
DeepSeek-V3.1: 40.3 (#144), Step 5 Preview: 51.8 (#36)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Coding | 1417 | 1480 |
| LMArena WebDev | — | 1564 |
| SciCode | — | 58.9% |
| WeirdML | 38.4% | — |
Reasoning Step 5 Preview leads
DeepSeek-V3.1: 27.9 (#110), Step 5 Preview: 40.0 (#57)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Hard Prompts | 1417 | 1465 |
| SimpleBench | 40% | — |
| Kagi LLM Benchmark | 53.2% | — |
| CritPt | — | 20.9% |
| DTBench | 82.7% | — |
| LMCA | 24.3% | — |
| Epoch Capabilities Index | 139.92 | — |
| ForecastBench | 58 | — |
Math Step 5 Preview leads
DeepSeek-V3.1: 38.9 (#122), Step 5 Preview: 46.1 (#72)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Math | 1420 | 1470 |
| ProofBench | — | 42% |
Knowledge DeepSeek-V3.1 leads
DeepSeek-V3.1: 43.7 (#90), Step 5 Preview: 41.2 (#112)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1405 | 1470 |
| Vectara Hallucination Rate | 5.5% | — |
Multimodal Not comparable
DeepSeek-V3.1: —, Step 5 Preview: 41.0 (#41)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Vision | — | 1267 |
Multilingual Step 5 Preview leads
DeepSeek-V3.1: 51.6 (#106), Step 5 Preview: 53.8 (#54)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1400 | 1432 |
| LMArena Chinese | 1469 | 1519 |
| LMArena Russian | 1405 | 1436 |
| LMArena Spanish | 1431 | 1439 |
| LMArena French | 1447 | — |
| LMArena German | 1411 | — |
| LMArena Japanese | 1378 | — |
| LMArena Korean | 1337 | — |
Instruction Following Step 5 Preview leads
DeepSeek-V3.1: 73.9 (#110), Step 5 Preview: 76.0 (#50)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1400 | 1444 |
Long Context Step 5 Preview leads
DeepSeek-V3.1: 36.3 (#232), Step 5 Preview: 44.5 (#64)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1422 | 1455 |
| Fiction.LiveBench | 52.8% | — |
Writing & Preference Step 5 Preview leads
DeepSeek-V3.1: 60.3 (#98), Step 5 Preview: 62.8 (#71)
| Benchmark | DeepSeek-V3.1 | Step 5 Preview |
|---|---|---|
| LMArena Text | 1420 | 1442 |
| LMArena Creative Writing | 1401 | 1410 |
| LMArena Multi-Turn | 1408 | 1447 |
| EQ-Bench Creative Writing | 1436 | — |
Frequently asked questions
Is DeepSeek-V3.1 better than Step 5 Preview?
Step 5 Preview is the stronger model overall, scoring 47.9 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 3.4× less per token, which makes it the better buy when Step 5 Preview's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.1 or Step 5 Preview?
DeepSeek-V3.1 is cheaper. It lists at $0.25 per million input tokens and $0.95 per million output tokens; Step 5 Preview lists at $1 and $2.70.
Is DeepSeek-V3.1 or Step 5 Preview better for coding?
Step 5 Preview scores higher on coding benchmarks: 51.8 versus 40.3 in the Noometry coding category.
Which has the bigger context window?
Step 5 Preview does, with 1.02M tokens against 164K.
How many benchmarks do DeepSeek-V3.1 and Step 5 Preview share?
13 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Step 5 Preview has 18.