Model comparison
GPT-5.6 Luna vs o3
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 47.5 on the Noometry Index.
Last verified . 35 shared benchmarks.
Summary
- They share 35 benchmarks with published results for both. GPT-5.6 Luna scores higher in 8 categories and o3 in 2 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.7 to 50.2.
- The biggest single-benchmark swing is ARC-AGI-2: 59.5% for GPT-5.6 Luna and 6.5% for o3.
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $2 / $8 for o3.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 200K.
Side by side
| GPT-5.6 Luna | o3 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 54.6 | 47.5 |
| Released | 2026-07-09 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 200K |
| Max output | 128K | 100K |
| Input $ / M tokens | $0.20 | $2 |
| Output $ / M tokens | $1.20 | $8 |
| Results tracked | 52 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Luna leads
GPT-5.6 Luna: 54.5 (#28), o3: 46.8 (#64)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| WeirdML | 60.9% | 52.4% |
| LMArena Coding | 1466 | 1408 |
| ALE-Bench | 1,667 | 933.55 |
| SWE-bench Verified | — | 62.3% |
| DeepSWE | 67.2% | — |
| FrontierCode | 39.8% | — |
| SWE-bench Verified (bash only) | — | 58.4% |
| Aider Polyglot | — | 81.3% |
| CursorBench | 35.9% | — |
| LMArena WebDev | 1519 | — |
| SciCode | 53.6% | — |
| GSO | — | 8.8% |
| CadEval | — | 74% |
Agentic & Tool Use Too close to call
GPT-5.6 Luna: 34.4 (#45), o3: 34.5 (#44)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| APEX-Agents | 43% | — |
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 30.8% |
| DeepResearch Bench | — | 45.2% |
| OSWorld | — | 23% |
| BALROG | 45.6% | — |
| GDP.pdf | 22.7% | — |
| LMArena Search | — | 1144 |
| METR Time Horizons | — | 65.4% |
| Vending-Bench 2 | 4,095 | — |
Reasoning GPT-5.6 Luna leads
GPT-5.6 Luna: 47.6 (#43), o3: 32.0 (#78)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| ARC-AGI-2 | 59.5% | 6.5% |
| SimpleBench | 46.8% | 53.1% |
| Kagi LLM Benchmark | 49.1% | 67.6% |
| ARC-AGI-1 | 88% | 60.8% |
| CritPt | 20.6% | 1.4% |
| Chess Puzzles | 40% | 38% |
| LMArena Hard Prompts | 1451 | 1402 |
| Mystery Game Puzzles | 21% | 29% |
| DTBench | 89.1% | 84.8% |
| LMCA | 48.5% | 39.7% |
| Epoch Capabilities Index | 156.39 | 146.86 |
| NYT Connections (extended) | 69.4% | — |
| EnigmaEval | — | 13.1% |
| Surface Evolver Bench | 61.9% | — |
| ForecastBench | — | 62.5 |
Math GPT-5.6 Luna leads
GPT-5.6 Luna: 77.7 (#14), o3: 50.2 (#58)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 82.1% | 33.3% |
| OTIS Mock AIME 2024-2025 | 98.3% | 84.4% |
| LMArena Math | 1458 | 1426 |
| FrontierMath Tier 4 | 61% | — |
| ProofBench | 60% | — |
| Omni-MATH | — | 71.4% |
| MATH Level 5 | — | 97.8% |
| FrontierMath (Feb 2025 set) | — | 18.7% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge GPT-5.6 Luna leads
GPT-5.6 Luna: 58.5 (#34), o3: 54.6 (#52)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| GPQA Diamond | 91.6% | 81.8% |
| SimpleQA Verified | 41% | 49.4% |
| LMArena Expert | 1478 | 1402 |
| Humanity's Last Exam | — | 20.3% |
| MMLU-Pro | — | 85.9% |
| Confabulations | — | 14.4% |
| GPQA (HELM) | — | 75.3% |
Multimodal GPT-5.6 Luna leads
GPT-5.6 Luna: 42.7 (#28), o3: 41.4 (#36)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| LMArena Vision | 1258 | 1214 |
| GeoBench | — | 74% |
| VPCT | — | 52% |
| Blueprint-Bench 2 | 22.6% | — |
| Furniture Assembly | 42.5% | — |
| LMArena Document | 1457 | — |
Multilingual GPT-5.6 Luna leads
GPT-5.6 Luna: 52.8 (#78), o3: 51.7 (#105)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| LMArena Non-English | 1417 | 1401 |
| LMArena Chinese | 1470 | 1437 |
| LMArena French | 1456 | 1430 |
| LMArena German | 1454 | 1420 |
| LMArena Japanese | 1411 | 1403 |
| LMArena Korean | 1415 | 1370 |
| LMArena Russian | 1428 | 1406 |
| LMArena Spanish | 1448 | 1395 |
Instruction Following GPT-5.6 Luna leads
GPT-5.6 Luna: 75.6 (#57), o3: 72.8 (#127)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| LMArena Instruction Following | 1437 | 1368 |
| IFEval | — | 86.9% |
Long Context o3 leads
GPT-5.6 Luna: 43.9 (#82), o3: 53.3 (#6)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| LMArena Longer Query | 1436 | 1372 |
| Fiction.LiveBench | — | 88.9% |
| CL-bench | — | 17.8% |
Writing & Preference GPT-5.6 Luna leads
GPT-5.6 Luna: 68.0 (#29), o3: 63.5 (#64)
| Benchmark | GPT-5.6 Luna | o3 |
|---|---|---|
| LMArena Text | 1431 | 1410 |
| LMArena Creative Writing | 1396 | 1359 |
| EQ-Bench Creative Writing | 1829 | 1676 |
| LMArena Multi-Turn | 1434 | 1405 |
| Short-Story Creative Writing | — | 83.9% |
| WildBench | — | 86.1% |
| EQ-Bench 4 | 1156 | — |
Frequently asked questions
Is GPT-5.6 Luna better than o3?
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 47.5 on the Noometry Index.
Which is cheaper, GPT-5.6 Luna or o3?
GPT-5.6 Luna is cheaper. It lists at $0.20 per million input tokens and $1.20 per million output tokens; o3 lists at $2 and $8.
Is GPT-5.6 Luna or o3 better for coding?
GPT-5.6 Luna scores higher on coding benchmarks: 54.5 versus 46.8 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 o3 share?
35 benchmarks have published results for both models. GPT-5.6 Luna has 52 scored results on Noometry and o3 has 63.