Model comparison
Gemini 3 Flash Preview vs GPT-5.6 Luna
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 52.3 on the Noometry Index.
Last verified . 41 shared benchmarks.
GoogleSummary
- They share 41 benchmarks with published results for both. Gemini 3 Flash Preview scores higher in 7 categories and GPT-5.6 Luna in 3 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.7 to 51.7.
- The biggest single-benchmark swing is ProofBench: 15% for Gemini 3 Flash Preview and 60% for GPT-5.6 Luna.
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $0.50 / $3 for Gemini 3 Flash Preview.
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 1.05M.
Side by side
| Gemini 3 Flash Preview | GPT-5.6 Luna | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 52.3 | 54.6 |
| Released | 2025-12-17 | 2026-07-09 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1.05M |
| Max output | 66K | 128K |
| Input $ / M tokens | $0.50 | $0.20 |
| Output $ / M tokens | $3 | $1.20 |
| Results tracked | 59 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Luna leads
Gemini 3 Flash Preview: 50.9 (#42), GPT-5.6 Luna: 54.5 (#28)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| LMArena WebDev | 1439 | 1519 |
| WeirdML | 61.6% | 60.9% |
| LMArena Coding | 1460 | 1466 |
| ALE-Bench | 1,367 | 1,667 |
| SWE-bench Verified | 75.4% | — |
| DeepSWE | — | 67.2% |
| FrontierCode | — | 39.8% |
| SWE-bench Verified (bash only) | 75.8% | — |
| CursorBench | — | 35.9% |
| SWE-bench Multilingual | 72.7% | — |
| SciCode | — | 53.6% |
| GSO | 9.8% | — |
Agentic & Tool Use Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 38.7 (#29), GPT-5.6 Luna: 34.4 (#45)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| BALROG | 48.1% | 45.6% |
| GDP.pdf | 10% | 22.7% |
| Vending-Bench 2 | 3,635 | 4,095 |
| Terminal-Bench | 64.3% | — |
| APEX-Agents | — | 43% |
| τ²-bench Airline | 82.5% | — |
| τ²-bench Banking | 27.3% | — |
| τ²-bench Retail | 76.8% | — |
| τ²-bench Telecom | 91.2% | — |
| DeepResearch Bench | 49.8% | — |
| LMArena Search | 1198 | — |
Reasoning Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 49.2 (#37), GPT-5.6 Luna: 47.6 (#43)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| ARC-AGI-2 | 33.6% | 59.5% |
| SimpleBench | 61.1% | 46.8% |
| NYT Connections (extended) | 83.1% | 69.4% |
| ARC-AGI-1 | 84.7% | 88% |
| Chess Puzzles | 40% | 40% |
| LMArena Hard Prompts | 1465 | 1451 |
| Mystery Game Puzzles | 26% | 21% |
| DTBench | 89.1% | 89.1% |
| LMCA | 43.1% | 48.5% |
| Epoch Capabilities Index | 151.8 | 156.39 |
| Kagi LLM Benchmark | — | 49.1% |
| CritPt | — | 20.6% |
| Surface Evolver Bench | — | 61.9% |
| ForecastBench | 58.5 | — |
Math GPT-5.6 Luna leads
Gemini 3 Flash Preview: 51.7 (#55), GPT-5.6 Luna: 77.7 (#14)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| FrontierMath (Tiers 1-3) | 51.2% | 82.1% |
| FrontierMath Tier 4 | 17.1% | 61% |
| OTIS Mock AIME 2024-2025 | 95.6% | 98.3% |
| ProofBench | 15% | 60% |
| LMArena Math | 1473 | 1458 |
| MathArena Final-Answer Competitions | 67.6% | — |
| FrontierMath (Feb 2025 set) | 35.6% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |
Knowledge Too close to call
Gemini 3 Flash Preview: 58.8 (#33), GPT-5.6 Luna: 58.5 (#34)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| GPQA Diamond | 89.4% | 91.6% |
| SimpleQA Verified | 66.8% | 41% |
| LMArena Expert | 1462 | 1478 |
| Vectara Hallucination Rate | 13.5% | — |
Multimodal Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 45.5 (#16), GPT-5.6 Luna: 42.7 (#28)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| LMArena Vision | 1285 | 1258 |
| Blueprint-Bench 2 | 0% | 22.6% |
| LMArena Document | 1413 | 1457 |
| GeoBench | 88% | — |
| VPCT | 72.6% | — |
| Furniture Assembly | — | 42.5% |
Multilingual Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 55.7 (#27), GPT-5.6 Luna: 52.8 (#78)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| LMArena Non-English | 1458 | 1417 |
| LMArena Chinese | 1511 | 1470 |
| LMArena French | 1477 | 1456 |
| LMArena German | 1497 | 1454 |
| LMArena Japanese | 1489 | 1411 |
| LMArena Korean | 1443 | 1415 |
| LMArena Russian | 1480 | 1428 |
| LMArena Spanish | 1469 | 1448 |
Instruction Following Too close to call
Gemini 3 Flash Preview: 75.7 (#56), GPT-5.6 Luna: 75.6 (#57)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| LMArena Instruction Following | 1437 | 1437 |
Long Context Too close to call
Gemini 3 Flash Preview: 44.4 (#67), GPT-5.6 Luna: 43.9 (#82)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| LMArena Longer Query | 1452 | 1436 |
Writing & Preference GPT-5.6 Luna leads
Gemini 3 Flash Preview: 65.5 (#45), GPT-5.6 Luna: 68.0 (#29)
| Benchmark | Gemini 3 Flash Preview | GPT-5.6 Luna |
|---|---|---|
| LMArena Text | 1466 | 1431 |
| LMArena Creative Writing | 1457 | 1396 |
| LMArena Multi-Turn | 1471 | 1434 |
| EQ-Bench Creative Writing | — | 1829 |
| EQ-Bench 4 | — | 1156 |
Frequently asked questions
Is Gemini 3 Flash Preview better than GPT-5.6 Luna?
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 52.3 on the Noometry Index.
Which is cheaper, Gemini 3 Flash Preview or GPT-5.6 Luna?
GPT-5.6 Luna is cheaper. It lists at $0.20 per million input tokens and $1.20 per million output tokens; Gemini 3 Flash Preview lists at $0.50 and $3.
Is Gemini 3 Flash Preview or GPT-5.6 Luna better for coding?
GPT-5.6 Luna scores higher on coding benchmarks: 54.5 versus 50.9 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Luna does, with 1.05M tokens against 1.05M.
How many benchmarks do Gemini 3 Flash Preview and GPT-5.6 Luna share?
41 benchmarks have published results for both models. Gemini 3 Flash Preview has 59 scored results on Noometry and GPT-5.6 Luna has 52.