Model comparison
DeepSeek-V3.2-Exp vs Step 5 Preview
Step 5 Preview is the stronger model overall, scoring 47.9 to 44.3 on the Noometry Index. DeepSeek-V3.2-Exp costs 4.9× less per token, which makes it the better buy when Step 5 Preview's lead doesn't matter for your workload.
Last verified . 17 shared benchmarks.
DeepSeek
StepFunSummary
- They share 17 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 2 categories and Step 5 Preview in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Step 5 Preview leads 40.0 to 22.1.
- The biggest single-benchmark swing is ProofBench: 8% for DeepSeek-V3.2-Exp and 42% for Step 5 Preview.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 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-V3.2-Exp has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.2-Exp | Step 5 Preview | |
|---|---|---|
| Provider | DeepSeek | StepFun |
| Noometry Index | 44.3 | 47.9 |
| Released | 2025-09-29 | 2026-09-16 |
| Weights | Open | Proprietary |
| Context window | 164K | 1.02M |
| Max output | 66K | 66K |
| Input $ / M tokens | $0.26 | $1 |
| Output $ / M tokens | $0.38 | $2.70 |
| Results tracked | 49 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 5 Preview leads
DeepSeek-V3.2-Exp: 46.5 (#65), Step 5 Preview: 51.8 (#36)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| LMArena WebDev | 1362 | 1564 |
| SciCode | 38.9% | 58.9% |
| LMArena Coding | 1454 | 1480 |
| SWE-bench Verified (bash only) | 70% | — |
| Aider Polyglot | 74.2% | — |
| SWE-bench Multilingual | 59% | — |
| WeirdML | 39.5% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3.2-Exp: 32.7 (#59), Step 5 Preview: —
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| Terminal-Bench | 39.6% | — |
| APEX-Agents | 21.3% | — |
| Berkeley Function Calling Leaderboard | 56.7% | — |
| TheAgentCompany | 42.9% | — |
| Vending-Bench 2 | 1,034 | — |
Reasoning Step 5 Preview leads
DeepSeek-V3.2-Exp: 22.1 (#208), Step 5 Preview: 40.0 (#57)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| CritPt | 2.9% | 20.9% |
| LMArena Hard Prompts | 1434 | 1465 |
| ARC-AGI-2 | 4% | — |
| Kagi LLM Benchmark | 52.2% | — |
| NYT Connections (extended) | 36.7% | — |
| ARC-AGI-1 | 57% | — |
| Chess Puzzles | 14% | — |
| Thematic Generalization | 65% | — |
| DTBench | 87.7% | — |
| LMCA | 29.1% | — |
| Epoch Capabilities Index | 146.27 | — |
Math Step 5 Preview leads
DeepSeek-V3.2-Exp: 41.7 (#87), Step 5 Preview: 46.1 (#72)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| ProofBench | 8% | 42% |
| LMArena Math | 1435 | 1470 |
| MathArena Final-Answer Competitions | 57.7% | — |
| OTIS Mock AIME 2024-2025 | 87.8% | — |
| FrontierMath (Feb 2025 set) | 22.1% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.7 (#66), Step 5 Preview: 41.2 (#112)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1436 | 1470 |
| GPQA Diamond | 83.4% | — |
| Vectara Hallucination Rate | 5.3% | — |
Multimodal Not comparable
DeepSeek-V3.2-Exp: —, Step 5 Preview: 41.0 (#41)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| LMArena Vision | — | 1267 |
Multilingual Step 5 Preview leads
DeepSeek-V3.2-Exp: 52.2 (#90), Step 5 Preview: 53.8 (#54)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1409 | 1432 |
| LMArena Chinese | 1461 | 1519 |
| LMArena Russian | 1424 | 1436 |
| LMArena Spanish | 1440 | 1439 |
| LMArena French | 1433 | — |
| LMArena German | 1440 | — |
| LMArena Japanese | 1374 | — |
| LMArena Korean | 1371 | — |
Instruction Following Step 5 Preview leads
DeepSeek-V3.2-Exp: 74.5 (#93), Step 5 Preview: 76.0 (#50)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1413 | 1444 |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#16), Step 5 Preview: 44.5 (#64)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1428 | 1455 |
| Fiction.LiveBench | 83.3% | — |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference Too close to call
DeepSeek-V3.2-Exp: 62.4 (#77), Step 5 Preview: 62.8 (#71)
| Benchmark | DeepSeek-V3.2-Exp | Step 5 Preview |
|---|---|---|
| LMArena Text | 1425 | 1442 |
| LMArena Creative Writing | 1403 | 1410 |
| LMArena Multi-Turn | 1427 | 1447 |
| EQ-Bench Creative Writing | 1515 | — |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than Step 5 Preview?
Step 5 Preview is the stronger model overall, scoring 47.9 to 44.3 on the Noometry Index. DeepSeek-V3.2-Exp costs 4.9× 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-V3.2-Exp or Step 5 Preview?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Step 5 Preview lists at $1 and $2.70.
Is DeepSeek-V3.2-Exp or Step 5 Preview better for coding?
Step 5 Preview scores higher on coding benchmarks: 51.8 versus 46.5 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-V3.2-Exp and Step 5 Preview share?
17 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and Step 5 Preview has 18.