Model comparison
Claude Sonnet 4 vs Step 5 Preview
Step 5 Preview is the stronger model overall, scoring 47.9 to 40.8 on the Noometry Index.
Last verified . 16 shared benchmarks.
StepFunSummary
- They share 16 benchmarks with published results for both. Claude Sonnet 4 scores higher in 1 category and Step 5 Preview in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Step 5 Preview leads 40.0 to 22.9.
- The biggest single-benchmark swing is CritPt: 0.3% for Claude Sonnet 4 and 20.9% for Step 5 Preview.
- Step 5 Preview is cheaper at $1 / $2.70 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- Step 5 Preview accepts more context: 1.02M tokens versus 200K.
Side by side
| Claude Sonnet 4 | Step 5 Preview | |
|---|---|---|
| Provider | Anthropic | StepFun |
| Noometry Index | 40.8 | 47.9 |
| Released | 2025-05-22 | 2026-09-16 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1.02M |
| Max output | 64K | 66K |
| Input $ / M tokens | $3 | $1 |
| Output $ / M tokens | $15 | $2.70 |
| Results tracked | 58 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 5 Preview leads
Claude Sonnet 4: 43.5 (#88), Step 5 Preview: 51.8 (#36)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| SciCode | 40% | 58.9% |
| LMArena Coding | 1414 | 1480 |
| SWE-bench Verified (bash only) | 64.9% | — |
| Aider Polyglot | 61.3% | — |
| LMArena WebDev | — | 1564 |
| GSO | 4.9% | — |
| WeirdML | 46.1% | — |
| ALE-Bench | 655.35 | — |
Agentic & Tool Use Not comparable
Claude Sonnet 4: 38.5 (#31), Step 5 Preview: —
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| METR Time Horizons | 62% | — |
Reasoning Step 5 Preview leads
Claude Sonnet 4: 22.9 (#187), Step 5 Preview: 40.0 (#57)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| CritPt | 0.3% | 20.9% |
| LMArena Hard Prompts | 1372 | 1465 |
| ARC-AGI-2 | 5.9% | — |
| SimpleBench | 45.5% | — |
| Kagi LLM Benchmark | 73% | — |
| ARC-AGI-1 | 40% | — |
| EnigmaEval | 3.1% | — |
| DTBench | 77.1% | — |
| LMCA | 29% | — |
| Epoch Capabilities Index | 141.69 | — |
| ForecastBench | 60.2 | — |
Math Step 5 Preview leads
Claude Sonnet 4: 43.3 (#80), Step 5 Preview: 46.1 (#72)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| LMArena Math | 1375 | 1470 |
| OTIS Mock AIME 2024-2025 | 71.1% | — |
| ProofBench | — | 42% |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge Too close to call
Claude Sonnet 4: 41.8 (#108), Step 5 Preview: 41.2 (#112)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1372 | 1470 |
| GPQA Diamond | 79.2% | — |
| Humanity's Last Exam | 7.8% | — |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| Vectara Hallucination Rate | 10.3% | — |
| GPQA (HELM) | 70.6% | — |
Multimodal Step 5 Preview leads
Claude Sonnet 4: 26.2 (#121), Step 5 Preview: 41.0 (#41)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| LMArena Vision | 1191 | 1267 |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |
Multilingual Step 5 Preview leads
Claude Sonnet 4: 46.7 (#156), Step 5 Preview: 53.8 (#54)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1333 | 1432 |
| LMArena Chinese | 1350 | 1519 |
| LMArena Russian | 1355 | 1436 |
| LMArena Spanish | 1357 | 1439 |
| LMArena French | 1363 | — |
| LMArena German | 1331 | — |
| LMArena Japanese | 1302 | — |
| LMArena Korean | 1291 | — |
Instruction Following Step 5 Preview leads
Claude Sonnet 4: 71.7 (#145), Step 5 Preview: 76.0 (#50)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1376 | 1444 |
| IFEval | 84% | — |
Long Context Step 5 Preview leads
Claude Sonnet 4: 33.7 (#259), Step 5 Preview: 44.5 (#64)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1398 | 1455 |
| Fiction.LiveBench | 46.9% | — |
Writing & Preference Step 5 Preview leads
Claude Sonnet 4: 57.1 (#132), Step 5 Preview: 62.8 (#71)
| Benchmark | Claude Sonnet 4 | Step 5 Preview |
|---|---|---|
| LMArena Text | 1351 | 1442 |
| LMArena Creative Writing | 1345 | 1410 |
| LMArena Multi-Turn | 1376 | 1447 |
| Short-Story Creative Writing | 81.4% | — |
| EQ-Bench Creative Writing | 1483 | — |
| WildBench | 83.8% | — |
Frequently asked questions
Is Claude Sonnet 4 better than Step 5 Preview?
Step 5 Preview is the stronger model overall, scoring 47.9 to 40.8 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 or Step 5 Preview?
Step 5 Preview is cheaper. It lists at $1 per million input tokens and $2.70 per million output tokens; Claude Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or Step 5 Preview better for coding?
Step 5 Preview scores higher on coding benchmarks: 51.8 versus 43.5 in the Noometry coding category.
Which has the bigger context window?
Step 5 Preview does, with 1.02M tokens against 200K.
How many benchmarks do Claude Sonnet 4 and Step 5 Preview share?
16 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and Step 5 Preview has 18.