# Claude Sonnet 4.6 vs GPT-5 Mini

> Claude Sonnet 4.6 is the stronger model overall, scoring 50.1 to 41.7 on the Noometry Index. GPT-5 Mini costs 8.7× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/claude-sonnet-4-6-vs-gpt-5-mini
- Last updated: 2026-10-10
- Shared benchmarks: 41

## Summary

- They share 41 benchmarks with published results for both. Claude Sonnet 4.6 scores higher in 10 categories and GPT-5 Mini in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Sonnet 4.6 leads 46.1 to 24.1.
- The biggest single-benchmark swing is ARC-AGI-2: 60.4% for Claude Sonnet 4.6 and 4.4% for GPT-5 Mini.
- GPT-5 Mini is cheaper at $0.25 / $2 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 Mini |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 50.1 | 41.7 |
| Rank | 42 | 115 |
| Context | 1M | 400K |
| Input $/M | $3 | $0.25 |
| Output $/M | $15 | $2 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Sonnet 4.6: 46.0 (#62)
- GPT-5 Mini: 40.0 (#134)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| SWE-bench Verified | 75.2% | 64.7% |
| SciCode | 46.8% | 39.2% |
| WeirdML | 66.1% | 52.7% |
| LMArena Coding | 1504 | 1406 |
| ALE-Bench | 1,327 | 799.77 |
| DeepSWE | 29.9% | — |
| FrontierCode | 24.3% | — |
| SWE-bench Verified (bash only) | — | 59.8% |
| LMArena WebDev | 1521 | — |
| SWE-bench Multilingual | — | 39.7% |
| AlgoTune | — | 1.38 |

## Agentic & Tool Use

- Claude Sonnet 4.6: 39.1 (#26)
- GPT-5 Mini: 31.0 (#64)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| Terminal-Bench | 53.4% | 34.8% |
| Vending-Bench 2 | 7,204 | -31.18 |
| APEX-Agents | 43% | — |
| Berkeley Function Calling Leaderboard | — | 55.5% |
| OSWorld 2.0 | 9.3% | — |
| DeepResearch Bench | 54.9% | — |
| OSWorld | 72.1% | — |
| ExploitBench | 23.6% | — |
| GBAEval | 48.8% | — |
| GDP.pdf | 18% | — |
| LMArena Search | 1221 | — |

## Reasoning

- Claude Sonnet 4.6: 46.1 (#40)
- GPT-5 Mini: 24.1 (#153)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| ARC-AGI-2 | 60.4% | 4.4% |
| ARC-AGI-1 | 86.5% | 54.3% |
| CritPt | 3.1% | 0% |
| Chess Puzzles | 13% | 30% |
| LMArena Hard Prompts | 1485 | 1380 |
| Mystery Game Puzzles | 16% | 10% |
| DTBench | 89.9% | 80.5% |
| LMCA | 46.5% | 34.2% |
| Epoch Capabilities Index | 152.25 | 145.51 |
| ForecastBench | 62 | 61 |
| Kagi LLM Benchmark | — | 70.3% |
| NYT Connections (extended) | 80.9% | — |
| EnigmaEval | — | 8.2% |
| Thematic Generalization | 76.3% | — |

## Math

- Claude Sonnet 4.6: 51.9 (#41)
- GPT-5 Mini: 46.3 (#59)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 85.8% | 86.7% |
| ProofBench | 45% | 9% |
| LMArena Math | 1463 | 1377 |
| FrontierMath (Feb 2025 set) | 32.4% | 27.2% |
| FrontierMath Tier 4 (v1) | 8.3% | 6.3% |
| FrontierMath (Tiers 1-3) | — | 46.7% |
| FrontierMath Tier 4 | — | 12.2% |
| Omni-MATH | — | 72.2% |
| MATH Level 5 | — | 97.8% |

## Knowledge

- Claude Sonnet 4.6: 51.6 (#60)
- GPT-5 Mini: 45.5 (#80)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| GPQA Diamond | 87.4% | 75% |
| SimpleQA Verified | 35.5% | 21.6% |
| Vectara Hallucination Rate | 10.6% | 12.9% |
| LMArena Expert | 1500 | 1377 |
| Humanity's Last Exam | — | 19.4% |
| MMLU-Pro | — | 83.5% |
| Confabulations | — | 13.3% |
| GPQA (HELM) | — | 75.6% |

## Multimodal

- Claude Sonnet 4.6: 38.0 (#61)
- GPT-5 Mini: 35.0 (#80)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| LMArena Vision | 1283 | 1202 |
| VPCT | — | 40.2% |
| Blueprint-Bench 2 | 6.7% | — |
| LMArena Document | 1482 | — |

## Multilingual

- Claude Sonnet 4.6: 54.6 (#39)
- GPT-5 Mini: 49.1 (#131)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| LMArena Non-English | 1440 | 1363 |
| LMArena Chinese | 1491 | 1387 |
| LMArena French | 1465 | 1385 |
| LMArena German | 1428 | 1366 |
| LMArena Japanese | 1419 | 1342 |
| LMArena Korean | 1411 | 1306 |
| LMArena Russian | 1440 | 1362 |
| LMArena Spanish | 1465 | 1355 |

## Instruction Following

- Claude Sonnet 4.6: 77.6 (#23)
- GPT-5 Mini: 76.4 (#43)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| LMArena Instruction Following | 1476 | 1357 |
| IFEval | — | 92.7% |

## Long Context

- Claude Sonnet 4.6: 45.3 (#40)
- GPT-5 Mini: 41.9 (#127)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| LMArena Longer Query | 1479 | 1355 |
| Fiction.LiveBench | — | 69.4% |

## Writing & Preference

- Claude Sonnet 4.6: 70.4 (#22)
- GPT-5 Mini: 55.4 (#143)

| Benchmark | Claude Sonnet 4.6 | GPT-5 Mini |
|---|---|---|
| LMArena Text | 1458 | 1373 |
| LMArena Creative Writing | 1435 | 1325 |
| EQ-Bench Creative Writing | 1810 | 1313 |
| LMArena Multi-Turn | 1464 | 1363 |
| Short-Story Creative Writing | — | 83.1% |
| WildBench | — | 85.5% |
| EQ-Bench 4 | 1207 | — |

## FAQ

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

Claude Sonnet 4.6 is the stronger model overall, scoring 50.1 to 41.7 on the Noometry Index. GPT-5 Mini costs 8.7× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.

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

GPT-5 Mini is cheaper. It lists at $0.25 per million input tokens and $2 per million output tokens; Claude Sonnet 4.6 lists at $3 and $15.

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

Claude Sonnet 4.6 scores higher on coding benchmarks: 46.0 versus 40.0 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 Mini share?

41 benchmarks have published results for both models. Claude Sonnet 4.6 has 57 scored results on Noometry and GPT-5 Mini has 60.