Model comparison
GPT-4o mini vs GPT-5.6 Luna
GPT-5.6 Luna is the stronger model overall, scoring 54.7 to 25.4 on the Noometry Index. GPT-4o mini costs 1.7× less per token, which makes it the better buy when GPT-5.6 Luna's lead doesn't matter for your workload.
Last verified . 33 shared benchmarks.
Summary
- They share 33 benchmarks with published results for both. GPT-4o mini scores higher in 0 categories and GPT-5.6 Luna in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.5 to 9.9.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.9% for GPT-4o mini and 98.3% for GPT-5.6 Luna.
- GPT-4o mini is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.20 / $1.20 for GPT-5.6 Luna.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 128K.
Side by side
| GPT-4o mini | GPT-5.6 Luna | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 25.4 | 54.7 |
| Released | 2024-07-18 | 2026-07-09 |
| Weights | Proprietary | Proprietary |
| Context window | 128K | 1.05M |
| Max output | 16K | 128K |
| Input $ / M tokens | $0.15 | $0.20 |
| Output $ / M tokens | $0.60 | $1.20 |
| Results tracked | 60 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Luna leads
GPT-4o mini: 21.6 (#330), GPT-5.6 Luna: 53.9 (#25)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| WeirdML | 11.8% | 60.9% |
| LMArena Coding | 1290 | 1466 |
| DeepSWE | — | 67.2% |
| FrontierCode | — | 39.8% |
| Aider Polyglot | 3.6% | — |
| CursorBench | — | 35.9% |
| LMArena WebDev | — | 1517 |
| SciCode | — | 53.6% |
| BigCodeBench Instruct | 46.1% | — |
| LiveBench Coding | 43.1% | — |
| BigCodeBench Complete | 57.4% | — |
| ALE-Bench | — | 1,667 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use GPT-5.6 Luna leads
GPT-4o mini: 27.2 (#95), GPT-5.6 Luna: 35.3 (#37)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| BALROG | 17.4% | 45.6% |
| GDP.pdf | — | 22.7% |
| Vending-Bench 2 | — | 4,095 |
Reasoning GPT-5.6 Luna leads
GPT-4o mini: 8.7 (#343), GPT-5.6 Luna: 47.5 (#38)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| ARC-AGI-2 | 0% | 59.5% |
| SimpleBench | 10.7% | 46.8% |
| Kagi LLM Benchmark | 28.8% | 49.1% |
| Chess Puzzles | 0% | 40% |
| LMArena Hard Prompts | 1267 | 1452 |
| Mystery Game Puzzles | 12% | 21% |
| DTBench | 54.4% | 89.1% |
| LMCA | 10.4% | 48.5% |
| Epoch Capabilities Index | 126.56 | 156.32 |
| NYT Connections (extended) | — | 69.4% |
| ARC-AGI-1 | — | 88% |
| CritPt | — | 20.6% |
| LiveBench Reasoning | 32.8% | — |
| LiveBench Data Analysis | 50% | — |
| Surface Evolver Bench | — | 61.9% |
| LiveBench | 41.3% | — |
| PIQA | 88.7% | — |
Math GPT-5.6 Luna leads
GPT-4o mini: 9.9 (#307), GPT-5.6 Luna: 77.5 (#12)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.7% | 82.1% |
| OTIS Mock AIME 2024-2025 | 6.9% | 98.3% |
| LMArena Math | 1267 | 1455 |
| FrontierMath Tier 4 | — | 61% |
| ProofBench | — | 60% |
| Omni-MATH | 28% | — |
| LiveBench Math | 36.3% | — |
| MATH Level 5 | 52.6% | — |
| GSM8K | 91.3% | — |
Knowledge GPT-5.6 Luna leads
GPT-4o mini: 17.7 (#279), GPT-5.6 Luna: 58.6 (#31)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| GPQA Diamond | 37.7% | 91.6% |
| SimpleQA Verified | 8.3% | 41% |
| LMArena Expert | 1235 | 1479 |
| MMLU-Pro | 60.3% | — |
| Confabulations | 37.2% | — |
| GPQA (HELM) | 36.8% | — |
| BoolQ | 88.7% | — |
| MMLU | 81.8% | — |
Multimodal GPT-5.6 Luna leads
GPT-4o mini: 25.3 (#117), GPT-5.6 Luna: 43.0 (#22)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| LMArena Vision | 1066 | 1259 |
| Video-MME | 64.8% | — |
| GeoBench | 64% | — |
| VPCT | 34% | — |
| Blueprint-Bench 2 | — | 22.6% |
| Furniture Assembly | — | 42.5% |
| LMArena Document | — | 1457 |
Multilingual GPT-5.6 Luna leads
GPT-4o mini: 42.2 (#194), GPT-5.6 Luna: 53.0 (#73)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| LMArena Non-English | 1266 | 1418 |
| LMArena Chinese | 1264 | 1469 |
| LMArena French | 1296 | 1457 |
| LMArena German | 1272 | 1459 |
| LMArena Japanese | 1216 | 1416 |
| LMArena Korean | 1195 | 1414 |
| LMArena Russian | 1275 | 1428 |
| LMArena Spanish | 1276 | 1445 |
Instruction Following GPT-5.6 Luna leads
GPT-4o mini: 62.1 (#234), GPT-5.6 Luna: 76.0 (#48)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| LMArena Instruction Following | 1259 | 1440 |
| LiveBench Instruction Following | 56.8% | — |
| IFEval | 78.2% | — |
Long Context GPT-5.6 Luna leads
GPT-4o mini: 39.1 (#181), GPT-5.6 Luna: 43.9 (#78)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| LMArena Longer Query | 1289 | 1437 |
Writing & Preference GPT-5.6 Luna leads
GPT-4o mini: 39.7 (#243), GPT-5.6 Luna: 68.3 (#27)
| Benchmark | GPT-4o mini | GPT-5.6 Luna |
|---|---|---|
| LMArena Text | 1286 | 1432 |
| LMArena Creative Writing | 1268 | 1397 |
| EQ-Bench Creative Writing | 873 | 1829 |
| LMArena Multi-Turn | 1285 | 1433 |
| Short-Story Creative Writing | 67.2% | — |
| WildBench | 79.1% | — |
| EQ-Bench 4 | — | 1156 |
| LiveBench Language | 28.6% | — |
Frequently asked questions
Is GPT-4o mini better than GPT-5.6 Luna?
GPT-5.6 Luna is the stronger model overall, scoring 54.7 to 25.4 on the Noometry Index. GPT-4o mini costs 1.7× less per token, which makes it the better buy when GPT-5.6 Luna's lead doesn't matter for your workload.
Which is cheaper, GPT-4o mini or GPT-5.6 Luna?
GPT-4o mini is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; GPT-5.6 Luna lists at $0.20 and $1.20.
Is GPT-4o mini or GPT-5.6 Luna better for coding?
GPT-5.6 Luna scores higher on coding benchmarks: 53.9 versus 21.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Luna does, with 1.05M tokens against 128K.
How many benchmarks do GPT-4o mini and GPT-5.6 Luna share?
33 benchmarks have published results for both models. GPT-4o mini has 60 scored results on Noometry and GPT-5.6 Luna has 51.