Model comparison
Gemini 3 Flash Preview vs GPT-6 Luna
GPT-6 Luna is the stronger model overall, scoring 53.3 to 52.3 on the Noometry Index.
Last verified . 36 shared benchmarks.
GoogleSummary
- They share 36 benchmarks with published results for both. Gemini 3 Flash Preview scores higher in 8 categories and GPT-6 Luna in 2 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Luna leads 76.1 to 51.7.
- The biggest single-benchmark swing is ProofBench: 15% for Gemini 3 Flash Preview and 64% for GPT-6 Luna.
- GPT-6 Luna is cheaper at $0.10 / $0.50 per million input/output tokens, against $0.50 / $3 for Gemini 3 Flash Preview.
- GPT-6 Luna accepts more context: 1.05M tokens versus 1.05M.
Side by side
| Gemini 3 Flash Preview | GPT-6 Luna | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 52.3 | 53.3 |
| Released | 2025-12-17 | 2026-09-22 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1.05M |
| Max output | 66K | 128K |
| Input $ / M tokens | $0.50 | $0.10 |
| Output $ / M tokens | $3 | $0.50 |
| Results tracked | 59 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Luna leads
Gemini 3 Flash Preview: 50.9 (#42), GPT-6 Luna: 55.5 (#25)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| LMArena WebDev | 1439 | 1581 |
| LMArena Coding | 1460 | 1439 |
| ALE-Bench | 1,367 | 1,577 |
| SWE-bench Verified | 75.4% | — |
| DeepSWE | — | 66.6% |
| FrontierCode | — | 42.4% |
| SWE-bench Verified (bash only) | 75.8% | — |
| SWE-bench Multilingual | 72.7% | — |
| SciCode | — | 54.6% |
| GSO | 9.8% | — |
| WeirdML | 61.6% | — |
Agentic & Tool Use Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 38.7 (#29), GPT-6 Luna: 33.3 (#54)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| GDP.pdf | 10% | 23% |
| Terminal-Bench | 64.3% | — |
| APEX-Agents | — | 44.3% |
| τ²-bench Airline | 82.5% | — |
| τ²-bench Banking | 27.3% | — |
| τ²-bench Retail | 76.8% | — |
| τ²-bench Telecom | 91.2% | — |
| DeepResearch Bench | 49.8% | — |
| BALROG | 48.1% | — |
| LMArena Search | 1198 | — |
| Vending-Bench 2 | 3,635 | — |
Reasoning Too close to call
Gemini 3 Flash Preview: 49.2 (#37), GPT-6 Luna: 48.2 (#41)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| ARC-AGI-2 | 33.6% | 59.3% |
| NYT Connections (extended) | 83.1% | 68.7% |
| ARC-AGI-1 | 84.7% | 86.7% |
| Chess Puzzles | 40% | 31% |
| LMArena Hard Prompts | 1465 | 1411 |
| Mystery Game Puzzles | 26% | 7% |
| DTBench | 89.1% | 90.1% |
| LMCA | 43.1% | 44.5% |
| Epoch Capabilities Index | 151.8 | 156.28 |
| SimpleBench | 61.1% | — |
| CritPt | — | 19.4% |
| ForecastBench | 58.5 | — |
Math GPT-6 Luna leads
Gemini 3 Flash Preview: 51.7 (#55), GPT-6 Luna: 76.1 (#15)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| FrontierMath (Tiers 1-3) | 51.2% | 78.9% |
| FrontierMath Tier 4 | 17.1% | 56.1% |
| OTIS Mock AIME 2024-2025 | 95.6% | 98.9% |
| ProofBench | 15% | 64% |
| LMArena Math | 1473 | 1416 |
| MathArena Final-Answer Competitions | 67.6% | — |
| FrontierMath (Feb 2025 set) | 35.6% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |
Knowledge Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 58.8 (#33), GPT-6 Luna: 57.0 (#41)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| GPQA Diamond | 89.4% | 90.5% |
| SimpleQA Verified | 66.8% | 41.4% |
| LMArena Expert | 1462 | 1444 |
| Vectara Hallucination Rate | 13.5% | — |
Multimodal Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 45.5 (#16), GPT-6 Luna: 42.4 (#30)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| LMArena Vision | 1285 | 1217 |
| Blueprint-Bench 2 | 0% | 31.2% |
| GeoBench | 88% | — |
| VPCT | 72.6% | — |
| Furniture Assembly | — | 44.2% |
| LMArena Document | 1413 | — |
Multilingual Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 55.7 (#27), GPT-6 Luna: 50.5 (#117)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| LMArena Non-English | 1458 | 1386 |
| LMArena Chinese | 1511 | 1433 |
| LMArena French | 1477 | 1420 |
| LMArena German | 1497 | 1369 |
| LMArena Japanese | 1489 | 1369 |
| LMArena Korean | 1443 | 1360 |
| LMArena Russian | 1480 | 1394 |
| LMArena Spanish | 1469 | 1393 |
Instruction Following Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 75.7 (#56), GPT-6 Luna: 74.3 (#99)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| LMArena Instruction Following | 1437 | 1409 |
Long Context Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 44.4 (#67), GPT-6 Luna: 43.0 (#111)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| LMArena Longer Query | 1452 | 1409 |
Writing & Preference Gemini 3 Flash Preview leads
Gemini 3 Flash Preview: 65.5 (#45), GPT-6 Luna: 58.3 (#119)
| Benchmark | Gemini 3 Flash Preview | GPT-6 Luna |
|---|---|---|
| LMArena Text | 1466 | 1391 |
| LMArena Creative Writing | 1457 | 1363 |
| LMArena Multi-Turn | 1471 | 1396 |
Frequently asked questions
Is Gemini 3 Flash Preview better than GPT-6 Luna?
GPT-6 Luna is the stronger model overall, scoring 53.3 to 52.3 on the Noometry Index.
Which is cheaper, Gemini 3 Flash Preview or GPT-6 Luna?
GPT-6 Luna is cheaper. It lists at $0.10 per million input tokens and $0.50 per million output tokens; Gemini 3 Flash Preview lists at $0.50 and $3.
Is Gemini 3 Flash Preview or GPT-6 Luna better for coding?
GPT-6 Luna scores higher on coding benchmarks: 55.5 versus 50.9 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Luna does, with 1.05M tokens against 1.05M.
How many benchmarks do Gemini 3 Flash Preview and GPT-6 Luna share?
36 benchmarks have published results for both models. Gemini 3 Flash Preview has 59 scored results on Noometry and GPT-6 Luna has 42.