# Claude Haiku 4.5 vs GPT-4

> Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 29.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-haiku-4-5-vs-gpt-4
- Last updated: 2026-10-12
- Shared benchmarks: 26

## Summary

- They share 26 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 7 categories and GPT-4 in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Haiku 4.5 leads 44.9 to 10.8.
- The biggest single-benchmark swing is MATH Level 5: 96.4% for Claude Haiku 4.5 and 23% for GPT-4.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $30 / $60 for GPT-4.
- Claude Haiku 4.5 accepts more context: 200K tokens versus 8K.

## Snapshot

| | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.5 | 29.1 |
| Rank | 165 | 316 |
| Context | 200K | 8K |
| Input $/M | $1 | $30 |
| Output $/M | $5 | $60 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Haiku 4.5: 44.0 (#78)
- GPT-4: 31.6 (#283)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| WeirdML | 45.4% | 12.4% |
| LMArena Coding | 1453 | 1254 |
| SWE-bench Verified (bash only) | 66.6% | — |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
| SciCode | 43.3% | — |
| BigCodeBench Instruct | — | 46% |
| BigCodeBench Complete | — | 57.2% |
| ALE-Bench | 653.48 | — |
| HumanEval+ | — | 79.3% |

## Agentic & Tool Use

- Claude Haiku 4.5: 33.6 (#52)
- GPT-4: —

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| Terminal-Bench | 35.5% | — |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| METR Time Horizons | — | 36.1% |
| Vending-Bench 2 | 458.89 | — |

## Reasoning

- Claude Haiku 4.5: 15.1 (#320)
- GPT-4: 17.8 (#289)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| Chess Puzzles | 8% | 4% |
| LMArena Hard Prompts | 1420 | 1241 |
| DTBench | 73.6% | 62.7% |
| LMCA | 30.9% | 17.1% |
| Epoch Capabilities Index | 142.41 | 125.89 |
| ForecastBench | 61.4 | 57.8 |
| ARC-AGI-2 | 4% | — |
| NYT Connections (extended) | 14.3% | — |
| ARC-AGI-1 | 47.7% | — |
| CritPt | 0% | — |
| Mystery Game Puzzles | — | 12% |
| BIG-Bench Hard | — | 75.1% |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |

## Math

- Claude Haiku 4.5: 44.9 (#78)
- GPT-4: 10.8 (#309)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 1.1% |
| LMArena Math | 1396 | 1269 |
| MATH Level 5 | 96.4% | 23% |
| Omni-MATH | 56.1% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
| GSM8K | — | 92% |

## Knowledge

- Claude Haiku 4.5: 37.7 (#153)
- GPT-4: 18.4 (#282)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| GPQA Diamond | 71.2% | 35.7% |
| LMArena Expert | 1442 | 1211 |
| SimpleQA Verified | 13.2% | — |
| MMLU-Pro | 77.7% | — |
| Vectara Hallucination Rate | 9.8% | — |
| GPQA (HELM) | 60.5% | — |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |

## Multimodal

- Claude Haiku 4.5: 26.8 (#118)
- GPT-4: —

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |

## Multilingual

- Claude Haiku 4.5: 49.9 (#129)
- GPT-4: 40.6 (#215)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| LMArena Non-English | 1377 | 1246 |
| LMArena Chinese | 1417 | 1242 |
| LMArena French | 1408 | 1283 |
| LMArena German | 1375 | 1251 |
| LMArena Japanese | 1339 | 1209 |
| LMArena Korean | 1347 | 1184 |
| LMArena Russian | 1381 | 1251 |
| LMArena Spanish | 1420 | 1261 |

## Instruction Following

- Claude Haiku 4.5: 71.4 (#149)
- GPT-4: 65.3 (#222)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1414 | 1241 |
| IFEval | 80.1% | — |

## Long Context

- Claude Haiku 4.5: 43.6 (#92)
- GPT-4: 37.7 (#212)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| LMArena Longer Query | 1427 | 1244 |

## Writing & Preference

- Claude Haiku 4.5: 57.9 (#123)
- GPT-4: 34.9 (#268)

| Benchmark | Claude Haiku 4.5 | GPT-4 |
|---|---|---|
| LMArena Text | 1396 | 1263 |
| LMArena Creative Writing | 1372 | 1244 |
| LMArena Multi-Turn | 1409 | 1257 |
| EQ-Bench Creative Writing | — | 752 |
| WildBench | 83.9% | — |
| EQ-Bench 4 | 1064 | — |

## FAQ

### Is Claude Haiku 4.5 better than GPT-4?

Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 29.1 on the Noometry Index.

### Which is cheaper, Claude Haiku 4.5 or GPT-4?

Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; GPT-4 lists at $30 and $60.

### Is Claude Haiku 4.5 or GPT-4 better for coding?

Claude Haiku 4.5 scores higher on coding benchmarks: 44.0 versus 31.6 in the Noometry coding category.

### Which has the bigger context window?

Claude Haiku 4.5 does, with 200K tokens against 8K.

### How many benchmarks do Claude Haiku 4.5 and GPT-4 share?

26 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and GPT-4 has 38.