# Claude Opus 5 vs GPT-5.6 Sol

> Claude Opus 5 is the stronger model overall, scoring 67.7 to 64.4 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-opus-5-vs-gpt-5-6-sol
- Last updated: 2026-10-10
- Shared benchmarks: 57

## Summary

- They share 57 benchmarks with published results for both. Claude Opus 5 scores higher in 10 categories and GPT-5.6 Sol in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in coding, where Claude Opus 5 leads 67.0 to 61.1.
- The biggest single-benchmark swing is GBAEval: 79.6% for Claude Opus 5 and 52.6% for GPT-5.6 Sol.
- GPT-5.6 Sol is cheaper at $4 / $20 per million input/output tokens, against $5 / $25 for Claude Opus 5.
- GPT-5.6 Sol accepts more context: 1.05M tokens versus 1M.

## Snapshot

| | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 67.7 | 64.4 |
| Rank | 4 | 6 |
| Context | 1M | 1.05M |
| Input $/M | $5 | $4 |
| Output $/M | $25 | $20 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Opus 5: 67.0 (#5)
- GPT-5.6 Sol: 61.1 (#7)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| DeepSWE | 73.6% | 72.7% |
| FrontierCode | 53.4% | 47.5% |
| CursorBench | 46.6% | 41.7% |
| LMArena WebDev | 1694 | 1619 |
| FrontierSWE | 52% | 32.2% |
| SciCode | 56.4% | 57.1% |
| WeirdML | 91.8% | 89.4% |
| LMArena Coding | 1533 | 1493 |
| ALE-Bench | 2,165 | 2,177 |
| MirrorCode | — | 20% |

## Agentic & Tool Use

- Claude Opus 5: 55.5 (#1)
- GPT-5.6 Sol: 50.2 (#7)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| APEX-Agents | 65.8% | 51.4% |
| OSWorld 2.0 | 31.4% | 27.3% |
| τ²-bench Banking | 48.7% | 46.9% |
| PostTrainBench | 35% | 36.2% |
| BALROG | 63.4% | 60% |
| GBAEval | 79.6% | 52.6% |
| GDP.pdf | 24% | 30.7% |
| Vending-Bench 2 | 11,182 | 9,619 |
| LMArena Search | — | 1257 |

## Reasoning

- Claude Opus 5: 77.1 (#3)
- GPT-5.6 Sol: 74.7 (#6)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| ARC-AGI-2 | 90.4% | 92.5% |
| SimpleBench | 80.6% | 71.7% |
| NYT Connections (extended) | 94.3% | 93.8% |
| ARC-AGI-1 | 97.5% | 97.5% |
| CritPt | 29.1% | 32.3% |
| Chess Puzzles | 42% | 64% |
| EBR-Bench | 45.7% | 44.8% |
| LMArena Hard Prompts | 1525 | 1481 |
| Mystery Game Puzzles | 59% | 58% |
| DTBench | 97.9% | 96% |
| LMCA | 64.5% | 59.2% |
| Bench to the Future 3 | 0.12 | 0.14 |
| Epoch Capabilities Index | 162.67 | 161.99 |
| Kagi LLM Benchmark | — | 67% |
| EnigmaEval | — | 37.1% |
| Surface Evolver Bench | — | 93.1% |

## Math

- Claude Opus 5: 86.1 (#6)
- GPT-5.6 Sol: 85.5 (#7)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| FrontierMath (Tiers 1-3) | 85.6% | 89.1% |
| FrontierMath Tier 4 | 73.2% | 82.9% |
| OTIS Mock AIME 2024-2025 | 98.9% | 100% |
| ProofBench | 99% | 83% |
| LMArena Math | 1533 | 1472 |
| FrontierMath Erdős | — | 0% |

## Knowledge

- Claude Opus 5: 66.8 (#8)
- GPT-5.6 Sol: 64.3 (#16)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| GPQA Diamond | 93.9% | 93.5% |
| SimpleQA Verified | 59.9% | 69.7% |
| LMArena Expert | 1558 | 1517 |
| Vectara Hallucination Rate | — | 12.4% |

## Multimodal

- Claude Opus 5: 51.1 (#6)
- GPT-5.6 Sol: 48.9 (#8)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| LMArena Vision | 1321 | 1281 |
| Blueprint-Bench 2 | 30.4% | 33.6% |
| Furniture Assembly | 60.8% | 56.7% |
| LMArena Document | 1516 | 1483 |

## Multilingual

- Claude Opus 5: 58.8 (#3)
- GPT-5.6 Sol: 55.4 (#32)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| LMArena Non-English | 1500 | 1450 |
| LMArena Chinese | 1575 | 1523 |
| LMArena French | 1522 | 1475 |
| LMArena German | 1522 | 1476 |
| LMArena Japanese | 1510 | 1475 |
| LMArena Korean | 1526 | 1434 |
| LMArena Russian | 1507 | 1466 |
| LMArena Spanish | 1519 | 1449 |

## Instruction Following

- Claude Opus 5: 79.3 (#5)
- GPT-5.6 Sol: 77.8 (#16)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| LMArena Instruction Following | 1516 | 1480 |

## Long Context

- Claude Opus 5: 46.5 (#20)
- GPT-5.6 Sol: 45.3 (#42)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| LMArena Longer Query | 1516 | 1479 |

## Writing & Preference

- Claude Opus 5: 79.3 (#2)
- GPT-5.6 Sol: 73.5 (#12)

| Benchmark | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| LMArena Text | 1506 | 1456 |
| LMArena Creative Writing | 1491 | 1450 |
| EQ-Bench Creative Writing | 2133 | 1972 |
| EQ-Bench 4 | 1385 | 1250 |
| LMArena Multi-Turn | 1496 | 1456 |

## FAQ

### Is Claude Opus 5 better than GPT-5.6 Sol?

Claude Opus 5 is the stronger model overall, scoring 67.7 to 64.4 on the Noometry Index.

### Which is cheaper, Claude Opus 5 or GPT-5.6 Sol?

GPT-5.6 Sol is cheaper. It lists at $4 per million input tokens and $20 per million output tokens; Claude Opus 5 lists at $5 and $25.

### Is Claude Opus 5 or GPT-5.6 Sol better for coding?

Claude Opus 5 scores higher on coding benchmarks: 67.0 versus 61.1 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.6 Sol does, with 1.05M tokens against 1M.

### How many benchmarks do Claude Opus 5 and GPT-5.6 Sol share?

57 benchmarks have published results for both models. Claude Opus 5 has 57 scored results on Noometry and GPT-5.6 Sol has 64.