# Claude Opus 4.5 vs GPT-4

> Claude Opus 4.5 is the stronger model overall, scoring 50.5 to 28.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-opus-4-5-vs-gpt-4
- Last updated: 2026-10-10
- Shared benchmarks: 28

## Summary

- They share 28 benchmarks with published results for both. Claude Opus 4.5 scores higher in 8 categories and GPT-4 in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Claude Opus 4.5 leads 56.4 to 18.4.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 86.1% for Claude Opus 4.5 and 1.1% for GPT-4.
- Claude Opus 4.5 is cheaper at $5 / $25 per million input/output tokens, against $30 / $60 for GPT-4.
- Claude Opus 4.5 accepts more context: 200K tokens versus 8K.

## Snapshot

| | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 50.5 | 28.9 |
| Rank | 39 | 308 |
| Context | 200K | 8K |
| Input $/M | $5 | $30 |
| Output $/M | $25 | $60 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Opus 4.5: 55.4 (#21)
- GPT-4: 31.2 (#277)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| WeirdML | 63.7% | 12.4% |
| LMArena Coding | 1504 | 1254 |
| SWE-bench Verified | 76.7% | — |
| SWE-bench Verified (bash only) | 76.8% | — |
| LMArena WebDev | 1493 | — |
| SWE-bench Multilingual | 70.7% | — |
| GSO | 26.5% | — |
| BigCodeBench Instruct | — | 46% |
| BigCodeBench Complete | — | 57.2% |
| ALE-Bench | 1,025 | — |
| AlgoTune | 1.77 | — |
| HumanEval+ | — | 79.3% |

## Agentic & Tool Use

- Claude Opus 4.5: 47.2 (#11)
- GPT-4: —

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| METR Time Horizons | 75% | 36.1% |
| Terminal-Bench | 63.1% | — |
| Berkeley Function Calling Leaderboard | 77.5% | — |
| GDPval | 45.5% | — |
| Remote Labor Index | 3.8% | — |
| τ²-bench Airline | 84% | — |
| τ²-bench Banking | 24.7% | — |
| τ²-bench Retail | 79.6% | — |
| τ²-bench Telecom | 92.3% | — |
| Cybench | 82% | — |
| DeepResearch Bench | 54.8% | — |
| OSWorld | 66.3% | — |
| BALROG | 43.5% | — |
| LMArena Search | 1180 | — |
| Vending-Bench 2 | 4,967 | — |

## Reasoning

- Claude Opus 4.5: 42.5 (#43)
- GPT-4: 17.3 (#285)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| Chess Puzzles | 12% | 4% |
| LMArena Hard Prompts | 1476 | 1241 |
| Mystery Game Puzzles | 22% | 12% |
| DTBench | 89.9% | 62.7% |
| LMCA | 44.5% | 17.1% |
| Epoch Capabilities Index | 150.1 | 125.88 |
| ForecastBench | 60.7 | 57.8 |
| ARC-AGI-2 | 37.6% | — |
| SimpleBench | 62% | — |
| Kagi LLM Benchmark | 80.2% | — |
| NYT Connections (extended) | 52.5% | — |
| ARC-AGI-1 | 80% | — |
| EnigmaEval | 11.9% | — |
| EBR-Bench | 14.3% | — |
| BIG-Bench Hard | — | 75.1% |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |

## Math

- Claude Opus 4.5: 38.5 (#85)
- GPT-4: 10.1 (#304)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 86.1% | 1.1% |
| LMArena Math | 1463 | 1269 |
| FrontierMath (Tiers 1-3) | 34.4% | — |
| FrontierMath Tier 4 | 4.9% | — |
| ProofBench | 36% | — |
| MATH Level 5 | — | 23% |
| FrontierMath (Feb 2025 set) | 20.7% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |
| GSM8K | — | 92% |

## Knowledge

- Claude Opus 4.5: 56.4 (#38)
- GPT-4: 18.4 (#277)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| GPQA Diamond | 86% | 35.7% |
| LMArena Expert | 1488 | 1211 |
| Humanity's Last Exam | 25.2% | — |
| SimpleQA Verified | 45.7% | — |
| Vectara Hallucination Rate | 10.9% | — |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |

## Multimodal

- Claude Opus 4.5: 31.1 (#99)
- GPT-4: —

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| GeoBench | 75% | — |
| VPCT | 40% | — |
| Furniture Assembly | 28.3% | — |
| LMArena Document | 1462 | — |

## Multilingual

- Claude Opus 4.5: 54.5 (#45)
- GPT-4: 40.8 (#210)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| LMArena Non-English | 1438 | 1246 |
| LMArena Chinese | 1470 | 1242 |
| LMArena French | 1471 | 1282 |
| LMArena German | 1449 | 1251 |
| LMArena Japanese | 1415 | 1209 |
| LMArena Korean | 1424 | 1184 |
| LMArena Russian | 1448 | 1251 |
| LMArena Spanish | 1455 | 1261 |

## Instruction Following

- Claude Opus 4.5: 77.7 (#20)
- GPT-4: 65.5 (#217)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1478 | 1241 |

## Long Context

- Claude Opus 4.5: 46.4 (#21)
- GPT-4: 37.7 (#207)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| LMArena Longer Query | 1480 | 1244 |
| CL-bench | 21.1% | — |

## Writing & Preference

- Claude Opus 4.5: 68.2 (#28)
- GPT-4: 35.0 (#263)

| Benchmark | Claude Opus 4.5 | GPT-4 |
|---|---|---|
| LMArena Text | 1451 | 1263 |
| LMArena Creative Writing | 1445 | 1244 |
| EQ-Bench Creative Writing | 1687 | 752 |
| LMArena Multi-Turn | 1466 | 1257 |

## FAQ

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

Claude Opus 4.5 is the stronger model overall, scoring 50.5 to 28.9 on the Noometry Index.

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

Claude Opus 4.5 is cheaper. It lists at $5 per million input tokens and $25 per million output tokens; GPT-4 lists at $30 and $60.

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

Claude Opus 4.5 scores higher on coding benchmarks: 55.4 versus 31.2 in the Noometry coding category.

### Which has the bigger context window?

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

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

28 benchmarks have published results for both models. Claude Opus 4.5 has 69 scored results on Noometry and GPT-4 has 38.