# GPT-5 Mini vs Grok 4

> Grok 4 is the stronger model overall, scoring 48.1 to 41.8 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gpt-5-mini-vs-grok-4
- Last updated: 2026-10-11
- Shared benchmarks: 39

## Summary

- They share 39 benchmarks with published results for both. GPT-5 Mini scores higher in 1 category and Grok 4 in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in long context, where Grok 4 leads 63.1 to 41.9.
- The biggest single-benchmark swing is Fiction.LiveBench: 69.4% for GPT-5 Mini and 94.4% for Grok 4.

## Snapshot

| | GPT-5 Mini | Grok 4 |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 41.8 | 48.1 |
| Rank | 128 | 56 |
| Context | 400K | — |
| Input $/M | $0.25 | — |
| Output $/M | $2 | — |
| Weights | Proprietary | Proprietary |

## Coding

- GPT-5 Mini: 40.1 (#146)
- Grok 4: 50.3 (#46)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| WeirdML | 52.7% | 45.7% |
| LMArena Coding | 1406 | 1408 |
| SWE-bench Verified | 64.7% | — |
| SWE-bench Verified (bash only) | 59.8% | — |
| Aider Polyglot | — | 79.6% |
| SWE-bench Multilingual | 39.7% | — |
| SciCode | 39.2% | — |
| ALE-Bench | 799.77 | — |
| AlgoTune | 1.38 | — |

## Agentic & Tool Use

- GPT-5 Mini: 31.1 (#70)
- Grok 4: 32.3 (#68)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| Terminal-Bench | 34.8% | 27.2% |
| Berkeley Function Calling Leaderboard | 55.5% | 63% |
| GDPval | — | 21.1% |
| Cybench | — | 43% |
| DeepResearch Bench | — | 47.3% |
| BALROG | — | 43.6% |
| LMArena Search | — | 1142 |
| METR Time Horizons | — | 66.6% |
| Vending-Bench 2 | -31.18 | — |

## Reasoning

- GPT-5 Mini: 23.9 (#168)
- Grok 4: 36.7 (#65)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| ARC-AGI-2 | 4.4% | 16% |
| Kagi LLM Benchmark | 70.3% | 73.6% |
| ARC-AGI-1 | 54.3% | 66.7% |
| Chess Puzzles | 30% | 28% |
| LMArena Hard Prompts | 1380 | 1409 |
| Epoch Capabilities Index | 145.52 | 146.44 |
| ForecastBench | 61 | 60.9 |
| SimpleBench | — | 60.5% |
| CritPt | 0% | — |
| EnigmaEval | 8.2% | — |
| Mystery Game Puzzles | 10% | — |
| DTBench | 80.5% | — |
| LMCA | 34.2% | — |

## Math

- GPT-5 Mini: 46.7 (#69)
- Grok 4: 48.4 (#64)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 86.7% | 84% |
| Omni-MATH | 72.2% | 60.3% |
| LMArena Math | 1378 | 1422 |
| FrontierMath (Feb 2025 set) | 27.2% | 19.7% |
| FrontierMath Tier 4 (v1) | 6.3% | 2.1% |
| FrontierMath (Tiers 1-3) | 46.7% | — |
| FrontierMath Tier 4 | 12.2% | — |
| ProofBench | 9% | — |
| MATH Level 5 | 97.8% | — |

## Knowledge

- GPT-5 Mini: 45.6 (#86)
- Grok 4: 53.8 (#55)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| GPQA Diamond | 75% | 87% |
| MMLU-Pro | 83.5% | 85.1% |
| Confabulations | 13.3% | 12.4% |
| GPQA (HELM) | 75.6% | 72.7% |
| LMArena Expert | 1379 | 1415 |
| Humanity's Last Exam | 19.4% | — |
| SimpleQA Verified | 21.6% | — |
| Vectara Hallucination Rate | 12.9% | — |

## Multimodal

- GPT-5 Mini: 35.6 (#85)
- Grok 4: 33.7 (#94)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| LMArena Vision | 1202 | 1210 |
| GeoBench | — | 45% |
| VPCT | 40.2% | — |

## Multilingual

- GPT-5 Mini: 48.9 (#137)
- Grok 4: 51.8 (#103)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| LMArena Non-English | 1363 | 1403 |
| LMArena Chinese | 1385 | 1427 |
| LMArena French | 1386 | 1418 |
| LMArena German | 1366 | 1429 |
| LMArena Japanese | 1341 | 1394 |
| LMArena Korean | 1308 | 1377 |
| LMArena Russian | 1362 | 1410 |
| LMArena Spanish | 1355 | 1420 |

## Instruction Following

- GPT-5 Mini: 76.2 (#46)
- Grok 4: 79.2 (#5)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| IFEval | 92.7% | 94.9% |
| LMArena Instruction Following | 1357 | 1387 |

## Long Context

- GPT-5 Mini: 41.9 (#132)
- Grok 4: 63.1 (#4)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| Fiction.LiveBench | 69.4% | 94.4% |
| LMArena Longer Query | 1355 | 1409 |

## Writing & Preference

- GPT-5 Mini: 55.2 (#148)
- Grok 4: 58.5 (#116)

| Benchmark | GPT-5 Mini | Grok 4 |
|---|---|---|
| LMArena Text | 1373 | 1411 |
| LMArena Creative Writing | 1325 | 1397 |
| Short-Story Creative Writing | 83.1% | 76.9% |
| WildBench | 85.5% | 79.7% |
| LMArena Multi-Turn | 1363 | 1416 |
| EQ-Bench Creative Writing | 1313 | — |

## FAQ

### Is GPT-5 Mini better than Grok 4?

Grok 4 is the stronger model overall, scoring 48.1 to 41.8 on the Noometry Index.

### Is GPT-5 Mini or Grok 4 better for coding?

Grok 4 scores higher on coding benchmarks: 50.3 versus 40.1 in the Noometry coding category.

### How many benchmarks do GPT-5 Mini and Grok 4 share?

39 benchmarks have published results for both models. GPT-5 Mini has 60 scored results on Noometry and Grok 4 has 48.