# GPT-5.6 Luna vs o4-mini

> GPT-5.6 Luna is the stronger model overall, scoring 54.7 to 41.6 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gpt-5-6-luna-vs-o4-mini
- Last updated: 2026-10-10
- Shared benchmarks: 35

## Summary

- They share 35 benchmarks with published results for both. GPT-5.6 Luna scores higher in 9 categories and o4-mini in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.5 to 40.2.
- The biggest single-benchmark swing is FrontierMath Tier 4: 61% for GPT-5.6 Luna and 4.9% for o4-mini.
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $1.10 / $4.40 for o4-mini.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 200K.

## Snapshot

| | GPT-5.6 Luna | o4-mini |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 54.7 | 41.6 |
| Rank | 24 | 119 |
| Context | 1.05M | 200K |
| Input $/M | $0.20 | $1.10 |
| Output $/M | $1.20 | $4.40 |
| Weights | Proprietary | Proprietary |

## Coding

- GPT-5.6 Luna: 53.9 (#25)
- o4-mini: 41.6 (#110)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| WeirdML | 60.9% | 52.6% |
| LMArena Coding | 1466 | 1369 |
| ALE-Bench | 1,667 | 826.17 |
| DeepSWE | 67.2% | — |
| FrontierCode | 39.8% | — |
| SWE-bench Verified (bash only) | — | 45% |
| Aider Polyglot | — | 72% |
| CursorBench | 35.9% | — |
| LMArena WebDev | 1517 | — |
| SciCode | 53.6% | — |
| GSO | — | 3.6% |
| CadEval | — | 62% |
| AlgoTune | — | 1.72 |

## Agentic & Tool Use

- GPT-5.6 Luna: 35.3 (#37)
- o4-mini: 32.3 (#57)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 53.2% |
| GDPval | — | 25.3% |
| BALROG | 45.6% | — |
| GDP.pdf | 22.7% | — |
| METR Time Horizons | — | 63.9% |
| Vending-Bench 2 | 4,095 | — |

## Reasoning

- GPT-5.6 Luna: 47.5 (#38)
- o4-mini: 24.6 (#143)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| ARC-AGI-2 | 59.5% | 6.1% |
| SimpleBench | 46.8% | 38.7% |
| Kagi LLM Benchmark | 49.1% | 67.6% |
| ARC-AGI-1 | 88% | 58.7% |
| CritPt | 20.6% | 0.6% |
| Chess Puzzles | 40% | 26% |
| LMArena Hard Prompts | 1452 | 1351 |
| Mystery Game Puzzles | 21% | 5% |
| DTBench | 89.1% | 77.6% |
| LMCA | 48.5% | 26.5% |
| Epoch Capabilities Index | 156.32 | 145.63 |
| NYT Connections (extended) | 69.4% | — |
| EnigmaEval | — | 9.2% |
| Surface Evolver Bench | 61.9% | — |
| ForecastBench | — | 61.8 |

## Math

- GPT-5.6 Luna: 77.5 (#12)
- o4-mini: 40.2 (#79)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 82.1% | 36.1% |
| FrontierMath Tier 4 | 61% | 4.9% |
| OTIS Mock AIME 2024-2025 | 98.3% | 81.7% |
| LMArena Math | 1455 | 1390 |
| ProofBench | 60% | — |
| Omni-MATH | — | 72% |
| MATH Level 5 | — | 97.8% |
| FrontierMath (Feb 2025 set) | — | 24.8% |
| FrontierMath Tier 4 (v1) | — | 6.3% |

## Knowledge

- GPT-5.6 Luna: 58.6 (#31)
- o4-mini: 43.5 (#83)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| GPQA Diamond | 91.6% | 79.6% |
| SimpleQA Verified | 41% | 19.6% |
| LMArena Expert | 1479 | 1344 |
| Humanity's Last Exam | — | 18.1% |
| MMLU-Pro | — | 82% |
| Confabulations | — | 15.8% |
| Vectara Hallucination Rate | — | 18.6% |
| GPQA (HELM) | — | 73.5% |

## Multimodal

- GPT-5.6 Luna: 43.0 (#22)
- o4-mini: 39.5 (#46)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| LMArena Vision | 1259 | 1194 |
| GeoBench | — | 64% |
| VPCT | — | 57.5% |
| Blueprint-Bench 2 | 22.6% | — |
| Furniture Assembly | 42.5% | — |
| LMArena Document | 1457 | — |

## Multilingual

- GPT-5.6 Luna: 53.0 (#73)
- o4-mini: 47.2 (#149)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| LMArena Non-English | 1418 | 1336 |
| LMArena Chinese | 1469 | 1355 |
| LMArena French | 1457 | 1362 |
| LMArena German | 1459 | 1336 |
| LMArena Japanese | 1416 | 1308 |
| LMArena Korean | 1414 | 1313 |
| LMArena Russian | 1428 | 1334 |
| LMArena Spanish | 1445 | 1346 |

## Instruction Following

- GPT-5.6 Luna: 76.0 (#48)
- o4-mini: 75.4 (#63)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| LMArena Instruction Following | 1440 | 1321 |
| IFEval | — | 92.8% |

## Long Context

- GPT-5.6 Luna: 43.9 (#78)
- o4-mini: 45.5 (#31)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| LMArena Longer Query | 1437 | 1315 |
| Fiction.LiveBench | — | 77.8% |

## Writing & Preference

- GPT-5.6 Luna: 68.3 (#27)
- o4-mini: 54.2 (#148)

| Benchmark | GPT-5.6 Luna | o4-mini |
|---|---|---|
| LMArena Text | 1432 | 1353 |
| LMArena Creative Writing | 1397 | 1294 |
| LMArena Multi-Turn | 1433 | 1350 |
| Short-Story Creative Writing | — | 75% |
| EQ-Bench Creative Writing | 1829 | — |
| WildBench | — | 85.4% |
| EQ-Bench 4 | 1156 | — |

## FAQ

### Is GPT-5.6 Luna better than o4-mini?

GPT-5.6 Luna is the stronger model overall, scoring 54.7 to 41.6 on the Noometry Index.

### Which is cheaper, GPT-5.6 Luna or o4-mini?

GPT-5.6 Luna is cheaper. It lists at $0.20 per million input tokens and $1.20 per million output tokens; o4-mini lists at $1.10 and $4.40.

### Is GPT-5.6 Luna or o4-mini better for coding?

GPT-5.6 Luna scores higher on coding benchmarks: 53.9 versus 41.6 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.6 Luna does, with 1.05M tokens against 200K.

### How many benchmarks do GPT-5.6 Luna and o4-mini share?

35 benchmarks have published results for both models. GPT-5.6 Luna has 51 scored results on Noometry and o4-mini has 60.