# Claude 3.5 Haiku vs Step 5 Preview

> Step 5 Preview is the stronger model overall, scoring 43.8 to 28.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-3-5-haiku-vs-step-5-preview
- Last updated: 2026-10-10
- Shared benchmarks: 2

## Summary

- They share 2 benchmarks with published results for both. Claude 3.5 Haiku scores higher in 0 categories and Step 5 Preview in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Step 5 Preview leads 38.0 to 17.1.
- The biggest single-benchmark swing is SciCode: 27.4% for Claude 3.5 Haiku and 58.9% for Step 5 Preview.

## Snapshot

| | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| Provider | Anthropic | StepFun |
| Noometry Index | 28.9 | 43.8 |
| Rank | 309 | 78 |
| Context | — | 1.02M |
| Input $/M | — | $1 |
| Output $/M | — | $2.70 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude 3.5 Haiku: 32.6 (#258)
- Step 5 Preview: 50.6 (#40)

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| SciCode | 27.4% | 58.9% |
| Aider Polyglot | 28% | — |
| LMArena WebDev | — | 1569 |
| WeirdML | 30.7% | — |
| BigCodeBench Instruct | 46.1% | — |
| LiveBench Coding | 51.4% | — |
| LMArena Coding | 1287 | — |
| BigCodeBench Complete | 59% | — |
| CadEval | 32% | — |

## Agentic & Tool Use

- Claude 3.5 Haiku: 27.7 (#90)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| BALROG | 19.3% | — |

## Reasoning

- Claude 3.5 Haiku: 17.1 (#286)
- Step 5 Preview: 38.0 (#58)

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| CritPt | 0% | 20.9% |
| LiveBench Reasoning | 28.1% | — |
| LMArena Hard Prompts | 1251 | — |
| DTBench | 56.7% | — |
| LiveBench Data Analysis | 48.5% | — |
| Epoch Capabilities Index | 127.15 | — |
| LiveBench | 43.5% | — |

## Math

- Claude 3.5 Haiku: 13.7 (#294)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 4.3% | — |
| Omni-MATH | 22.4% | — |
| LiveBench Math | 35.5% | — |
| LMArena Math | 1244 | — |
| MATH Level 5 | 46.4% | — |
| FrontierMath (Feb 2025 set) | 0.3% | — |

## Knowledge

- Claude 3.5 Haiku: 18.7 (#276)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| GPQA Diamond | 38.1% | — |
| MMLU-Pro | 60.5% | — |
| Confabulations | 36.7% | — |
| GPQA (HELM) | 36.3% | — |
| LMArena Expert | 1208 | — |
| MMLU | 74.3% | — |

## Multimodal

- Claude 3.5 Haiku: 26.3 (#113)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| LMArena Vision | 1092 | — |
| GeoBench | 34% | — |

## Multilingual

- Claude 3.5 Haiku: 40.2 (#213)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1238 | — |
| LMArena Chinese | 1229 | — |
| LMArena French | 1262 | — |
| LMArena German | 1237 | — |
| LMArena Japanese | 1175 | — |
| LMArena Korean | 1172 | — |
| LMArena Russian | 1253 | — |
| LMArena Spanish | 1261 | — |

## Instruction Following

- Claude 3.5 Haiku: 63.1 (#229)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| LiveBench Instruction Following | 61.9% | — |
| IFEval | 79.2% | — |
| LMArena Instruction Following | 1241 | — |

## Long Context

- Claude 3.5 Haiku: 38.3 (#195)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1261 | — |

## Writing & Preference

- Claude 3.5 Haiku: 42.9 (#228)
- Step 5 Preview: —

| Benchmark | Claude 3.5 Haiku | Step 5 Preview |
|---|---|---|
| LMArena Text | 1255 | — |
| LMArena Creative Writing | 1233 | — |
| Short-Story Creative Writing | 73.5% | — |
| EQ-Bench Creative Writing | 1146 | — |
| WildBench | 76% | — |
| LMArena Multi-Turn | 1265 | — |
| LiveBench Language | 35.4% | — |

## FAQ

### Is Claude 3.5 Haiku better than Step 5 Preview?

Step 5 Preview is the stronger model overall, scoring 43.8 to 28.9 on the Noometry Index.

### Is Claude 3.5 Haiku or Step 5 Preview better for coding?

Step 5 Preview scores higher on coding benchmarks: 50.6 versus 32.6 in the Noometry coding category.

### How many benchmarks do Claude 3.5 Haiku and Step 5 Preview share?

2 benchmarks have published results for both models. Claude 3.5 Haiku has 49 scored results on Noometry and Step 5 Preview has 3.