# Claude Haiku 4.5 vs Step 3.5 Flash

> Step 3.5 Flash is the stronger model overall, scoring 41.7 to 39.3 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-haiku-4-5-vs-step-3-5-flash
- Last updated: 2026-10-10
- Shared benchmarks: 18

## Summary

- They share 18 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 3 categories and Step 3.5 Flash in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Step 3.5 Flash leads 21.2 to 15.1.
- The biggest single-benchmark swing is NYT Connections (extended): 14.3% for Claude Haiku 4.5 and 28.4% for Step 3.5 Flash.
- Step 3.5 Flash is cheaper at $0.10 / $0.30 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
- Step 3.5 Flash accepts more context: 256K tokens versus 200K.
- Step 3.5 Flash has downloadable open weights; the other is API-only.

## Snapshot

| | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| Provider | Anthropic | StepFun |
| Noometry Index | 39.3 | 41.7 |
| Rank | 155 | 116 |
| Context | 200K | 256K |
| Input $/M | $1 | $0.10 |
| Output $/M | $5 | $0.30 |
| Weights | Proprietary | Open |

## Coding

- Claude Haiku 4.5: 43.9 (#71)
- Step 3.5 Flash: 41.7 (#106)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| LMArena Coding | 1454 | 1436 |
| SWE-bench Verified (bash only) | 66.6% | — |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
| SciCode | 43.3% | — |
| WeirdML | 45.4% | — |
| ALE-Bench | 653.48 | — |

## Agentic & Tool Use

- Claude Haiku 4.5: 33.5 (#50)
- Step 3.5 Flash: —

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| Terminal-Bench | 35.5% | — |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| Vending-Bench 2 | 458.89 | — |

## Reasoning

- Claude Haiku 4.5: 15.1 (#310)
- Step 3.5 Flash: 21.2 (#202)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| NYT Connections (extended) | 14.3% | 28.4% |
| LMArena Hard Prompts | 1420 | 1411 |
| ARC-AGI-2 | 4% | — |
| ARC-AGI-1 | 47.7% | — |
| CritPt | 0% | — |
| Chess Puzzles | 8% | — |
| DTBench | 73.6% | — |
| LMCA | 30.9% | — |
| Epoch Capabilities Index | 142.42 | — |
| ForecastBench | 61.4 | — |

## Math

- Claude Haiku 4.5: 43.0 (#69)
- Step 3.5 Flash: 40.6 (#77)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| LMArena Math | 1395 | 1408 |
| MathArena Final-Answer Competitions | — | 66.8% |
| OTIS Mock AIME 2024-2025 | 66.7% | — |
| Omni-MATH | 56.1% | — |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |

## Knowledge

- Claude Haiku 4.5: 37.6 (#146)
- Step 3.5 Flash: 39.0 (#127)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| LMArena Expert | 1443 | 1422 |
| GPQA Diamond | 71.2% | — |
| SimpleQA Verified | 13.2% | — |
| MMLU-Pro | 77.7% | — |
| Vectara Hallucination Rate | 9.8% | — |
| GPQA (HELM) | 60.5% | — |

## Multimodal

- Claude Haiku 4.5: 27.0 (#111)
- Step 3.5 Flash: —

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |

## Multilingual

- Claude Haiku 4.5: 50.1 (#124)
- Step 3.5 Flash: 50.7 (#115)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| LMArena Non-English | 1376 | 1386 |
| LMArena Chinese | 1415 | 1448 |
| LMArena French | 1408 | 1421 |
| LMArena German | 1377 | 1404 |
| LMArena Japanese | 1339 | 1353 |
| LMArena Korean | 1346 | 1350 |
| LMArena Russian | 1380 | 1385 |
| LMArena Spanish | 1420 | 1419 |

## Instruction Following

- Claude Haiku 4.5: 71.6 (#144)
- Step 3.5 Flash: 73.3 (#119)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| LMArena Instruction Following | 1414 | 1385 |
| IFEval | 80.1% | — |

## Long Context

- Claude Haiku 4.5: 43.6 (#86)
- Step 3.5 Flash: 42.8 (#112)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| LMArena Longer Query | 1427 | 1402 |

## Writing & Preference

- Claude Haiku 4.5: 58.1 (#119)
- Step 3.5 Flash: 59.0 (#108)

| Benchmark | Claude Haiku 4.5 | Step 3.5 Flash |
|---|---|---|
| LMArena Text | 1396 | 1403 |
| LMArena Creative Writing | 1371 | 1357 |
| LMArena Multi-Turn | 1409 | 1405 |
| WildBench | 83.9% | — |
| EQ-Bench 4 | 1064 | — |

## FAQ

### Is Claude Haiku 4.5 better than Step 3.5 Flash?

Step 3.5 Flash is the stronger model overall, scoring 41.7 to 39.3 on the Noometry Index.

### Which is cheaper, Claude Haiku 4.5 or Step 3.5 Flash?

Step 3.5 Flash is cheaper. It lists at $0.10 per million input tokens and $0.30 per million output tokens; Claude Haiku 4.5 lists at $1 and $5.

### Is Claude Haiku 4.5 or Step 3.5 Flash better for coding?

Claude Haiku 4.5 scores higher on coding benchmarks: 43.9 versus 41.7 in the Noometry coding category.

### Which has the bigger context window?

Step 3.5 Flash does, with 256K tokens against 200K.

### How many benchmarks do Claude Haiku 4.5 and Step 3.5 Flash share?

18 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and Step 3.5 Flash has 19.