# Claude Sonnet 4.6 vs GPT-5

> Claude Sonnet 4.6 and GPT-5 score almost the same on the Noometry Index (50.1 vs 50.9), so choose on price, context window or the category you care about most.

- Canonical page: https://noometry.com/compare/claude-sonnet-4-6-vs-gpt-5
- Last updated: 2026-10-10
- Shared benchmarks: 43

## Summary

- They share 43 benchmarks with published results for both. Claude Sonnet 4.6 scores higher in 5 categories and GPT-5 in 5 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 45.3.
- The biggest single-benchmark swing is ARC-AGI-2: 60.4% for Claude Sonnet 4.6 and 9.9% for GPT-5.
- GPT-5 is cheaper at $1.25 / $10 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.6.
- Claude Sonnet 4.6 accepts more context: 1M tokens versus 400K.

## Snapshot

| | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 50.1 | 50.9 |
| Rank | 42 | 37 |
| Context | 1M | 400K |
| Input $/M | $3 | $1.25 |
| Output $/M | $15 | $10 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Sonnet 4.6: 46.0 (#62)
- GPT-5: 50.7 (#39)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| SWE-bench Verified | 75.2% | 73.6% |
| LMArena WebDev | 1521 | 1417 |
| SciCode | 46.8% | 42.9% |
| WeirdML | 66.1% | 60.7% |
| LMArena Coding | 1504 | 1436 |
| ALE-Bench | 1,327 | 1,162 |
| DeepSWE | 29.9% | — |
| FrontierCode | 24.3% | — |
| SWE-bench Verified (bash only) | — | 65% |
| Aider Polyglot | — | 88% |
| GSO | — | 6.9% |
| AlgoTune | — | 1.67 |

## Agentic & Tool Use

- Claude Sonnet 4.6: 39.1 (#26)
- GPT-5: 33.0 (#53)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| Terminal-Bench | 53.4% | 49.6% |
| DeepResearch Bench | 54.9% | 49.6% |
| LMArena Search | 1221 | 1133 |
| APEX-Agents | 43% | — |
| OSWorld 2.0 | 9.3% | — |
| GDPval | — | 34.8% |
| Remote Labor Index | — | 1.7% |
| OSWorld | 72.1% | — |
| BALROG | — | 32.8% |
| ExploitBench | 23.6% | — |
| GBAEval | 48.8% | — |
| GDP.pdf | 18% | — |
| METR Time Horizons | — | 69.6% |
| Vending-Bench 2 | 7,204 | — |

## Reasoning

- Claude Sonnet 4.6: 46.1 (#40)
- GPT-5: 38.3 (#55)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| ARC-AGI-2 | 60.4% | 9.9% |
| ARC-AGI-1 | 86.5% | 65.7% |
| CritPt | 3.1% | 12.6% |
| Chess Puzzles | 13% | 37% |
| LMArena Hard Prompts | 1485 | 1415 |
| Mystery Game Puzzles | 16% | 23% |
| DTBench | 89.9% | 90.7% |
| LMCA | 46.5% | 40% |
| Epoch Capabilities Index | 152.25 | 150 |
| ForecastBench | 62 | 61.4 |
| SimpleBench | — | 56.7% |
| Kagi LLM Benchmark | — | 72.7% |
| NYT Connections (extended) | 80.9% | — |
| EnigmaEval | — | 10.5% |
| Thematic Generalization | 76.3% | — |
| EBR-Bench | — | 12.7% |

## Math

- Claude Sonnet 4.6: 51.9 (#41)
- GPT-5: 54.6 (#36)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 85.8% | 91.4% |
| ProofBench | 45% | 18% |
| LMArena Math | 1463 | 1406 |
| FrontierMath (Feb 2025 set) | 32.4% | 32.4% |
| FrontierMath Tier 4 (v1) | 8.3% | 12.5% |
| FrontierMath (Tiers 1-3) | — | 55.4% |
| FrontierMath Tier 4 | — | 22% |
| Omni-MATH | — | 64.7% |
| MATH Level 5 | — | 98.1% |

## Knowledge

- Claude Sonnet 4.6: 51.6 (#60)
- GPT-5: 56.5 (#37)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| GPQA Diamond | 87.4% | 86.2% |
| SimpleQA Verified | 35.5% | 50.1% |
| Vectara Hallucination Rate | 10.6% | 14.7% |
| LMArena Expert | 1500 | 1420 |
| Humanity's Last Exam | — | 25.3% |
| MMLU-Pro | — | 86.3% |
| Confabulations | — | 10.3% |
| GPQA (HELM) | — | 79.2% |

## Multimodal

- Claude Sonnet 4.6: 38.0 (#61)
- GPT-5: 46.1 (#12)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| LMArena Vision | 1283 | 1232 |
| GeoBench | — | 81% |
| VPCT | — | 66% |
| Blueprint-Bench 2 | 6.7% | — |
| LMArena Document | 1482 | — |

## Multilingual

- Claude Sonnet 4.6: 54.6 (#39)
- GPT-5: 51.5 (#105)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| LMArena Non-English | 1440 | 1397 |
| LMArena Chinese | 1491 | 1423 |
| LMArena French | 1465 | 1409 |
| LMArena German | 1428 | 1415 |
| LMArena Japanese | 1419 | 1411 |
| LMArena Korean | 1411 | 1361 |
| LMArena Russian | 1440 | 1406 |
| LMArena Spanish | 1465 | 1398 |

## Instruction Following

- Claude Sonnet 4.6: 77.6 (#23)
- GPT-5: 74.0 (#107)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| LMArena Instruction Following | 1476 | 1388 |
| IFEval | — | 87.5% |

## Long Context

- Claude Sonnet 4.6: 45.3 (#40)
- GPT-5: 69.5 (#2)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| LMArena Longer Query | 1479 | 1399 |
| Fiction.LiveBench | — | 97.2% |

## Writing & Preference

- Claude Sonnet 4.6: 70.4 (#22)
- GPT-5: 63.5 (#63)

| Benchmark | Claude Sonnet 4.6 | GPT-5 |
|---|---|---|
| LMArena Text | 1458 | 1406 |
| LMArena Creative Writing | 1435 | 1365 |
| EQ-Bench Creative Writing | 1810 | 1627 |
| LMArena Multi-Turn | 1464 | 1426 |
| Short-Story Creative Writing | — | 86% |
| WildBench | — | 85.7% |
| EQ-Bench 4 | 1207 | — |

## FAQ

### Is Claude Sonnet 4.6 better than GPT-5?

Claude Sonnet 4.6 and GPT-5 score almost the same on the Noometry Index (50.1 vs 50.9), so choose on price, context window or the category you care about most.

### Which is cheaper, Claude Sonnet 4.6 or GPT-5?

GPT-5 is cheaper. It lists at $1.25 per million input tokens and $10 per million output tokens; Claude Sonnet 4.6 lists at $3 and $15.

### Is Claude Sonnet 4.6 or GPT-5 better for coding?

GPT-5 scores higher on coding benchmarks: 50.7 versus 46.0 in the Noometry coding category.

### Which has the bigger context window?

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

### How many benchmarks do Claude Sonnet 4.6 and GPT-5 share?

43 benchmarks have published results for both models. Claude Sonnet 4.6 has 57 scored results on Noometry and GPT-5 has 69.