# Claude Sonnet 4 vs GPT-4o mini

> Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 25.4 on the Noometry Index. GPT-4o mini costs 23× less per token, which makes it the better buy when Claude Sonnet 4's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/claude-sonnet-4-vs-gpt-4o-mini
- Last updated: 2026-10-10
- Shared benchmarks: 39

## Summary

- They share 39 benchmarks with published results for both. Claude Sonnet 4 scores higher in 9 categories and GPT-4o mini in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Sonnet 4 leads 41.5 to 9.9.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 71.1% for Claude Sonnet 4 and 6.9% for GPT-4o mini.
- GPT-4o mini is cheaper at $0.15 / $0.60 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- Claude Sonnet 4 accepts more context: 200K tokens versus 128K.

## Snapshot

| | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 40.6 | 25.4 |
| Rank | 139 | 337 |
| Context | 200K | 128K |
| Input $/M | $3 | $0.15 |
| Output $/M | $15 | $0.60 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Sonnet 4: 44.0 (#70)
- GPT-4o mini: 21.6 (#330)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| Aider Polyglot | 61.3% | 3.6% |
| WeirdML | 46.1% | 11.8% |
| LMArena Coding | 1414 | 1290 |
| SWE-bench Verified (bash only) | 64.9% | — |
| SciCode | 40% | — |
| GSO | 4.9% | — |
| BigCodeBench Instruct | — | 46.1% |
| LiveBench Coding | — | 43.1% |
| BigCodeBench Complete | — | 57.4% |
| ALE-Bench | 655.35 | — |
| HumanEval+ | — | 83.5% |
| MBPP+ | — | 72.2% |

## Agentic & Tool Use

- Claude Sonnet 4: 38.3 (#29)
- GPT-4o mini: 27.2 (#95)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| BALROG | — | 17.4% |
| METR Time Horizons | 62% | — |

## Reasoning

- Claude Sonnet 4: 23.0 (#169)
- GPT-4o mini: 8.7 (#343)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| ARC-AGI-2 | 5.9% | 0% |
| SimpleBench | 45.5% | 10.7% |
| Kagi LLM Benchmark | 73% | 28.8% |
| LMArena Hard Prompts | 1372 | 1267 |
| DTBench | 77.1% | 54.4% |
| LMCA | 29% | 10.4% |
| Epoch Capabilities Index | 141.69 | 126.56 |
| ARC-AGI-1 | 40% | — |
| CritPt | 0.3% | — |
| Chess Puzzles | — | 0% |
| EnigmaEval | 3.1% | — |
| LiveBench Reasoning | — | 32.8% |
| Mystery Game Puzzles | — | 12% |
| LiveBench Data Analysis | — | 50% |
| ForecastBench | 60.2 | — |
| LiveBench | — | 41.3% |
| PIQA | — | 88.7% |

## Math

- Claude Sonnet 4: 41.5 (#71)
- GPT-4o mini: 9.9 (#307)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 6.9% |
| Omni-MATH | 60.2% | 28% |
| LMArena Math | 1375 | 1267 |
| MATH Level 5 | 84.4% | 52.6% |
| FrontierMath (Tiers 1-3) | — | 0.7% |
| LiveBench Math | — | 36.3% |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
| GSM8K | — | 91.3% |

## Knowledge

- Claude Sonnet 4: 41.6 (#99)
- GPT-4o mini: 17.7 (#279)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| GPQA Diamond | 79.2% | 37.7% |
| MMLU-Pro | 84.3% | 60.3% |
| Confabulations | 13.2% | 37.2% |
| GPQA (HELM) | 70.6% | 36.8% |
| LMArena Expert | 1371 | 1235 |
| Humanity's Last Exam | 7.8% | — |
| SimpleQA Verified | — | 8.3% |
| Vectara Hallucination Rate | 10.3% | — |
| BoolQ | — | 88.7% |
| MMLU | — | 81.8% |

## Multimodal

- Claude Sonnet 4: 25.6 (#116)
- GPT-4o mini: 25.3 (#117)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| LMArena Vision | 1191 | 1066 |
| GeoBench | 37% | 64% |
| VPCT | 34% | 34% |
| Video-MME | — | 64.8% |
| MindCube | 44.8% | — |

## Multilingual

- Claude Sonnet 4: 46.9 (#151)
- GPT-4o mini: 42.2 (#194)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| LMArena Non-English | 1333 | 1266 |
| LMArena Chinese | 1351 | 1264 |
| LMArena French | 1364 | 1296 |
| LMArena German | 1331 | 1272 |
| LMArena Japanese | 1301 | 1216 |
| LMArena Korean | 1293 | 1195 |
| LMArena Russian | 1355 | 1275 |
| LMArena Spanish | 1356 | 1276 |

## Instruction Following

- Claude Sonnet 4: 71.8 (#141)
- GPT-4o mini: 62.1 (#234)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| IFEval | 84% | 78.2% |
| LMArena Instruction Following | 1375 | 1259 |
| LiveBench Instruction Following | — | 56.8% |

## Long Context

- Claude Sonnet 4: 33.6 (#254)
- GPT-4o mini: 39.1 (#181)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| LMArena Longer Query | 1398 | 1289 |
| Fiction.LiveBench | 46.9% | — |

## Writing & Preference

- Claude Sonnet 4: 57.3 (#127)
- GPT-4o mini: 39.7 (#243)

| Benchmark | Claude Sonnet 4 | GPT-4o mini |
|---|---|---|
| LMArena Text | 1351 | 1286 |
| LMArena Creative Writing | 1345 | 1268 |
| Short-Story Creative Writing | 81.4% | 67.2% |
| EQ-Bench Creative Writing | 1483 | 873 |
| WildBench | 83.8% | 79.1% |
| LMArena Multi-Turn | 1376 | 1285 |
| LiveBench Language | — | 28.6% |

## FAQ

### Is Claude Sonnet 4 better than GPT-4o mini?

Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 25.4 on the Noometry Index. GPT-4o mini costs 23× less per token, which makes it the better buy when Claude Sonnet 4's lead doesn't matter for your workload.

### Which is cheaper, Claude Sonnet 4 or GPT-4o mini?

GPT-4o mini is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Claude Sonnet 4 lists at $3 and $15.

### Is Claude Sonnet 4 or GPT-4o mini better for coding?

Claude Sonnet 4 scores higher on coding benchmarks: 44.0 versus 21.6 in the Noometry coding category.

### Which has the bigger context window?

Claude Sonnet 4 does, with 200K tokens against 128K.

### How many benchmarks do Claude Sonnet 4 and GPT-4o mini share?

39 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and GPT-4o mini has 60.