# Claude Opus 4.6 vs GPT-5.2

> Claude Opus 4.6 is the stronger model overall, scoring 58.3 to 54.1 on the Noometry Index. GPT-5.2 costs 2.1× less per token, which makes it the better buy when Claude Opus 4.6's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/claude-opus-4-6-vs-gpt-5-2
- Last updated: 2026-10-10
- Shared benchmarks: 61

## Summary

- They share 61 benchmarks with published results for both. Claude Opus 4.6 scores higher in 9 categories and GPT-5.2 in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in multimodal, where GPT-5.2 leads 50.9 to 37.0.
- The biggest single-benchmark swing is ProofBench: 50% for Claude Opus 4.6 and 15% for GPT-5.2.
- GPT-5.2 is cheaper at $1.75 / $14 per million input/output tokens, against $5 / $25 for Claude Opus 4.6.
- Claude Opus 4.6 accepts more context: 1M tokens versus 400K.

## Snapshot

| | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 58.3 | 54.1 |
| Rank | 18 | 31 |
| Context | 1M | 400K |
| Input $/M | $5 | $1.75 |
| Output $/M | $25 | $14 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Opus 4.6: 57.7 (#15)
- GPT-5.2: 52.2 (#32)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| SWE-bench Verified | 78.7% | 73.8% |
| SWE-bench Verified (bash only) | 75.6% | 72.8% |
| LMArena WebDev | 1546 | 1415 |
| SWE-bench Multilingual | 72% | 66.7% |
| GSO | 41.2% | 27.4% |
| WeirdML | 78% | 72.2% |
| LMArena Coding | 1535 | 1447 |
| ALE-Bench | 996.5 | 1,294 |
| AlgoTune | 1.47 | 2.05 |
| FrontierCode | 26.6% | — |

## Agentic & Tool Use

- Claude Opus 4.6: 51.0 (#4)
- GPT-5.2: 40.0 (#24)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| Terminal-Bench | 79.8% | 64.9% |
| Remote Labor Index | 4.2% | 2.5% |
| τ²-bench Banking | 27.3% | 32.2% |
| DeepResearch Bench | 55.3% | 41.1% |
| LMArena Search | 1253 | 1207 |
| METR Time Horizons | 78.9% | 75.3% |
| Vending-Bench 2 | 8,018 | 3,591 |
| APEX-Agents | 46.3% | — |
| Berkeley Function Calling Leaderboard | — | 55.9% |
| GDPval | — | 49.7% |
| τ²-bench Airline | — | 83% |
| τ²-bench Retail | — | 81.6% |
| τ²-bench Telecom | — | 89.7% |
| Cybench | 93% | — |
| GBAEval | 44.1% | — |

## Reasoning

- Claude Opus 4.6: 57.7 (#21)
- GPT-5.2: 50.1 (#33)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 69.2% | 52.9% |
| SimpleBench | 67.6% | 45.8% |
| Kagi LLM Benchmark | 83.6% | 73.3% |
| NYT Connections (extended) | 92.1% | 83.6% |
| ARC-AGI-1 | 94% | 86.2% |
| Chess Puzzles | 17% | 49% |
| EnigmaEval | 7.6% | 10.4% |
| EBR-Bench | 12.7% | 23% |
| LMArena Hard Prompts | 1527 | 1445 |
| Mystery Game Puzzles | 25% | 23% |
| DTBench | 91.2% | 90.9% |
| LMCA | 55.8% | 43.9% |
| Epoch Capabilities Index | 155.36 | 153.45 |
| ForecastBench | 60 | 60.1 |
| Thematic Generalization | 80.6% | — |

## Math

- Claude Opus 4.6: 62.7 (#25)
- GPT-5.2: 59.8 (#32)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 66% | 67.4% |
| FrontierMath Tier 4 | 26.8% | 31.7% |
| MathArena Final-Answer Competitions | 78.5% | 72% |
| OTIS Mock AIME 2024-2025 | 94.4% | 96.1% |
| ProofBench | 50% | 15% |
| LMArena Math | 1517 | 1439 |
| FrontierMath (Feb 2025 set) | 40.7% | 40.7% |
| FrontierMath Tier 4 (v1) | 22.9% | 18.8% |

## Knowledge

- Claude Opus 4.6: 61.8 (#23)
- GPT-5.2: 59.2 (#29)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 90.5% | 91.4% |
| Humanity's Last Exam | 34.4% | 27.8% |
| SimpleQA Verified | 47% | 37.1% |
| Vectara Hallucination Rate | 12.2% | 8.4% |
| LMArena Expert | 1546 | 1445 |

## Multimodal

- Claude Opus 4.6: 37.0 (#67)
- GPT-5.2: 50.9 (#7)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| LMArena Vision | 1316 | 1268 |
| Furniture Assembly | 28.3% | 38.3% |
| LMArena Document | 1507 | 1405 |
| VPCT | — | 84% |

## Multilingual

- Claude Opus 4.6: 58.2 (#4)
- GPT-5.2: 53.6 (#63)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1490 | 1425 |
| LMArena Chinese | 1551 | 1460 |
| LMArena French | 1513 | 1456 |
| LMArena German | 1504 | 1448 |
| LMArena Japanese | 1484 | 1421 |
| LMArena Korean | 1465 | 1390 |
| LMArena Russian | 1497 | 1440 |
| LMArena Spanish | 1512 | 1433 |

## Instruction Following

- Claude Opus 4.6: 79.7 (#3)
- GPT-5.2: 74.9 (#83)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1523 | 1417 |

## Long Context

- Claude Opus 4.6: 48.1 (#13)
- GPT-5.2: 44.0 (#74)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| CL-bench | 20.7% | 18.2% |
| LMArena Longer Query | 1520 | 1428 |
| CL-bench Life | 17% | — |

## Writing & Preference

- Claude Opus 4.6: 73.7 (#9)
- GPT-5.2: 67.0 (#30)

| Benchmark | Claude Opus 4.6 | GPT-5.2 |
|---|---|---|
| LMArena Text | 1504 | 1439 |
| LMArena Creative Writing | 1506 | 1402 |
| EQ-Bench Creative Writing | 1809 | 1703 |
| LMArena Multi-Turn | 1514 | 1458 |
| EQ-Bench 4 | 1223 | — |

## FAQ

### Is Claude Opus 4.6 better than GPT-5.2?

Claude Opus 4.6 is the stronger model overall, scoring 58.3 to 54.1 on the Noometry Index. GPT-5.2 costs 2.1× less per token, which makes it the better buy when Claude Opus 4.6's lead doesn't matter for your workload.

### Which is cheaper, Claude Opus 4.6 or GPT-5.2?

GPT-5.2 is cheaper. It lists at $1.75 per million input tokens and $14 per million output tokens; Claude Opus 4.6 lists at $5 and $25.

### Is Claude Opus 4.6 or GPT-5.2 better for coding?

Claude Opus 4.6 scores higher on coding benchmarks: 57.7 versus 52.2 in the Noometry coding category.

### Which has the bigger context window?

Claude Opus 4.6 does, with 1M tokens against 400K.

### How many benchmarks do Claude Opus 4.6 and GPT-5.2 share?

61 benchmarks have published results for both models. Claude Opus 4.6 has 68 scored results on Noometry and GPT-5.2 has 67.