Model comparison
Claude Sonnet 4.6 vs Step 5 Preview
Claude Sonnet 4.6 is the stronger model overall, scoring 50.3 to 47.9 on the Noometry Index. Step 5 Preview costs 4.2× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
StepFunSummary
- They share 18 benchmarks with published results for both. Claude Sonnet 4.6 scores higher in 7 categories and Step 5 Preview in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Claude Sonnet 4.6 leads 51.7 to 41.2.
- The biggest single-benchmark swing is CritPt: 3.1% for Claude Sonnet 4.6 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 4.6.
- Step 5 Preview accepts more context: 1.02M tokens versus 1M.
Side by side
| Claude Sonnet 4.6 | Step 5 Preview | |
|---|---|---|
| Provider | Anthropic | StepFun |
| Noometry Index | 50.3 | 47.9 |
| Released | 2026-02-17 | 2026-09-16 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.02M |
| Max output | 128K | 66K |
| Input $ / M tokens | $3 | $1 |
| Output $ / M tokens | $15 | $2.70 |
| Results tracked | 57 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 5 Preview leads
Claude Sonnet 4.6: 46.3 (#67), Step 5 Preview: 51.8 (#36)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| LMArena WebDev | 1522 | 1564 |
| SciCode | 46.8% | 58.9% |
| LMArena Coding | 1504 | 1480 |
| SWE-bench Verified | 75.2% | — |
| DeepSWE | 29.9% | — |
| FrontierCode | 24.3% | — |
| WeirdML | 66.1% | — |
| ALE-Bench | 1,327 | — |
Agentic & Tool Use Not comparable
Claude Sonnet 4.6: 39.1 (#28), Step 5 Preview: —
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| Terminal-Bench | 53.4% | — |
| APEX-Agents | 43% | — |
| OSWorld 2.0 | 9.3% | — |
| DeepResearch Bench | 54.9% | — |
| OSWorld | 72.1% | — |
| ExploitBench | 23.6% | — |
| GBAEval | 48.8% | — |
| GDP.pdf | 18% | — |
| LMArena Search | 1221 | — |
| Vending-Bench 2 | 7,204 | — |
Reasoning Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 46.1 (#45), Step 5 Preview: 40.0 (#57)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| CritPt | 3.1% | 20.9% |
| LMArena Hard Prompts | 1484 | 1465 |
| ARC-AGI-2 | 60.4% | — |
| NYT Connections (extended) | 80.9% | — |
| ARC-AGI-1 | 86.5% | — |
| Chess Puzzles | 13% | — |
| Thematic Generalization | 76.3% | — |
| Mystery Game Puzzles | 16% | — |
| DTBench | 89.9% | — |
| LMCA | 46.5% | — |
| Epoch Capabilities Index | 152.24 | — |
| ForecastBench | 62 | — |
Math Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 52.9 (#49), Step 5 Preview: 46.1 (#72)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| ProofBench | 45% | 42% |
| LMArena Math | 1462 | 1470 |
| OTIS Mock AIME 2024-2025 | 85.8% | — |
| FrontierMath (Feb 2025 set) | 32.4% | — |
| FrontierMath Tier 4 (v1) | 8.3% | — |
Knowledge Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 51.7 (#65), Step 5 Preview: 41.2 (#112)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1500 | 1470 |
| GPQA Diamond | 87.4% | — |
| SimpleQA Verified | 35.5% | — |
| Vectara Hallucination Rate | 10.6% | — |
Multimodal Step 5 Preview leads
Claude Sonnet 4.6: 38.0 (#68), Step 5 Preview: 41.0 (#41)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| LMArena Vision | 1283 | 1267 |
| Blueprint-Bench 2 | 6.7% | — |
| LMArena Document | 1482 | — |
Multilingual Too close to call
Claude Sonnet 4.6: 54.4 (#41), Step 5 Preview: 53.8 (#54)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1440 | 1432 |
| LMArena Chinese | 1491 | 1519 |
| LMArena Russian | 1440 | 1436 |
| LMArena Spanish | 1464 | 1439 |
| LMArena French | 1465 | — |
| LMArena German | 1428 | — |
| LMArena Japanese | 1420 | — |
| LMArena Korean | 1411 | — |
Instruction Following Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 77.4 (#25), Step 5 Preview: 76.0 (#50)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1475 | 1444 |
Long Context Too close to call
Claude Sonnet 4.6: 45.3 (#44), Step 5 Preview: 44.5 (#64)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1479 | 1455 |
Writing & Preference Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 70.2 (#22), Step 5 Preview: 62.8 (#71)
| Benchmark | Claude Sonnet 4.6 | Step 5 Preview |
|---|---|---|
| LMArena Text | 1458 | 1442 |
| LMArena Creative Writing | 1435 | 1410 |
| LMArena Multi-Turn | 1464 | 1447 |
| EQ-Bench Creative Writing | 1810 | — |
| EQ-Bench 4 | 1207 | — |
Frequently asked questions
Is Claude Sonnet 4.6 better than Step 5 Preview?
Claude Sonnet 4.6 is the stronger model overall, scoring 50.3 to 47.9 on the Noometry Index. Step 5 Preview costs 4.2× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Which is cheaper, Claude Sonnet 4.6 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 4.6 lists at $3 and $15.
Is Claude Sonnet 4.6 or Step 5 Preview better for coding?
Step 5 Preview scores higher on coding benchmarks: 51.8 versus 46.3 in the Noometry coding category.
Which has the bigger context window?
Step 5 Preview does, with 1.02M tokens against 1M.
How many benchmarks do Claude Sonnet 4.6 and Step 5 Preview share?
18 benchmarks have published results for both models. Claude Sonnet 4.6 has 57 scored results on Noometry and Step 5 Preview has 18.