Model comparison
GPT-5 Mini vs GPT-6 Luna
GPT-6 Luna is the stronger model overall, scoring 53.3 to 41.8 on the Noometry Index.
Last verified . 34 shared benchmarks.
Summary
- They share 34 benchmarks with published results for both. GPT-5 Mini scores higher in 1 category and GPT-6 Luna in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Luna leads 76.1 to 46.7.
- The biggest single-benchmark swing is ProofBench: 9% for GPT-5 Mini and 64% for GPT-6 Luna.
- GPT-6 Luna is cheaper at $0.10 / $0.50 per million input/output tokens, against $0.25 / $2 for GPT-5 Mini.
- GPT-6 Luna accepts more context: 1.05M tokens versus 400K.
Side by side
| GPT-5 Mini | GPT-6 Luna | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 41.8 | 53.3 |
| Released | 2025-08-07 | 2026-09-22 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $0.25 | $0.10 |
| Output $ / M tokens | $2 | $0.50 |
| Results tracked | 60 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Luna leads
GPT-5 Mini: 40.1 (#146), GPT-6 Luna: 55.5 (#25)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| SciCode | 39.2% | 54.6% |
| LMArena Coding | 1406 | 1439 |
| ALE-Bench | 799.77 | 1,577 |
| SWE-bench Verified | 64.7% | — |
| DeepSWE | — | 66.6% |
| FrontierCode | — | 42.4% |
| SWE-bench Verified (bash only) | 59.8% | — |
| LMArena WebDev | — | 1581 |
| SWE-bench Multilingual | 39.7% | — |
| WeirdML | 52.7% | — |
| AlgoTune | 1.38 | — |
Agentic & Tool Use GPT-6 Luna leads
GPT-5 Mini: 31.1 (#70), GPT-6 Luna: 33.3 (#54)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| Terminal-Bench | 34.8% | — |
| APEX-Agents | — | 44.3% |
| Berkeley Function Calling Leaderboard | 55.5% | — |
| GDP.pdf | — | 23% |
| Vending-Bench 2 | -31.18 | — |
Reasoning GPT-6 Luna leads
GPT-5 Mini: 23.9 (#168), GPT-6 Luna: 48.2 (#41)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| ARC-AGI-2 | 4.4% | 59.3% |
| ARC-AGI-1 | 54.3% | 86.7% |
| CritPt | 0% | 19.4% |
| Chess Puzzles | 30% | 31% |
| LMArena Hard Prompts | 1380 | 1411 |
| Mystery Game Puzzles | 10% | 7% |
| DTBench | 80.5% | 90.1% |
| LMCA | 34.2% | 44.5% |
| Epoch Capabilities Index | 145.52 | 156.28 |
| Kagi LLM Benchmark | 70.3% | — |
| NYT Connections (extended) | — | 68.7% |
| EnigmaEval | 8.2% | — |
| ForecastBench | 61 | — |
Math GPT-6 Luna leads
GPT-5 Mini: 46.7 (#69), GPT-6 Luna: 76.1 (#15)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| FrontierMath (Tiers 1-3) | 46.7% | 78.9% |
| FrontierMath Tier 4 | 12.2% | 56.1% |
| OTIS Mock AIME 2024-2025 | 86.7% | 98.9% |
| ProofBench | 9% | 64% |
| LMArena Math | 1378 | 1416 |
| Omni-MATH | 72.2% | — |
| MATH Level 5 | 97.8% | — |
| FrontierMath (Feb 2025 set) | 27.2% | — |
| FrontierMath Tier 4 (v1) | 6.3% | — |
Knowledge GPT-6 Luna leads
GPT-5 Mini: 45.6 (#86), GPT-6 Luna: 57.0 (#41)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| GPQA Diamond | 75% | 90.5% |
| SimpleQA Verified | 21.6% | 41.4% |
| LMArena Expert | 1379 | 1444 |
| Humanity's Last Exam | 19.4% | — |
| MMLU-Pro | 83.5% | — |
| Confabulations | 13.3% | — |
| Vectara Hallucination Rate | 12.9% | — |
| GPQA (HELM) | 75.6% | — |
Multimodal GPT-6 Luna leads
GPT-5 Mini: 35.6 (#85), GPT-6 Luna: 42.4 (#30)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| LMArena Vision | 1202 | 1217 |
| VPCT | 40.2% | — |
| Blueprint-Bench 2 | — | 31.2% |
| Furniture Assembly | — | 44.2% |
Multilingual GPT-6 Luna leads
GPT-5 Mini: 48.9 (#137), GPT-6 Luna: 50.5 (#117)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| LMArena Non-English | 1363 | 1386 |
| LMArena Chinese | 1385 | 1433 |
| LMArena French | 1386 | 1420 |
| LMArena German | 1366 | 1369 |
| LMArena Japanese | 1341 | 1369 |
| LMArena Korean | 1308 | 1360 |
| LMArena Russian | 1362 | 1394 |
| LMArena Spanish | 1355 | 1393 |
Instruction Following GPT-5 Mini leads
GPT-5 Mini: 76.2 (#46), GPT-6 Luna: 74.3 (#99)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| LMArena Instruction Following | 1357 | 1409 |
| IFEval | 92.7% | — |
Long Context GPT-6 Luna leads
GPT-5 Mini: 41.9 (#132), GPT-6 Luna: 43.0 (#111)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| LMArena Longer Query | 1355 | 1409 |
| Fiction.LiveBench | 69.4% | — |
Writing & Preference GPT-6 Luna leads
GPT-5 Mini: 55.2 (#148), GPT-6 Luna: 58.3 (#119)
| Benchmark | GPT-5 Mini | GPT-6 Luna |
|---|---|---|
| LMArena Text | 1373 | 1391 |
| LMArena Creative Writing | 1325 | 1363 |
| LMArena Multi-Turn | 1363 | 1396 |
| Short-Story Creative Writing | 83.1% | — |
| EQ-Bench Creative Writing | 1313 | — |
| WildBench | 85.5% | — |
Frequently asked questions
Is GPT-5 Mini better than GPT-6 Luna?
GPT-6 Luna is the stronger model overall, scoring 53.3 to 41.8 on the Noometry Index.
Which is cheaper, GPT-5 Mini 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; GPT-5 Mini lists at $0.25 and $2.
Is GPT-5 Mini or GPT-6 Luna better for coding?
GPT-6 Luna scores higher on coding benchmarks: 55.5 versus 40.1 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Luna does, with 1.05M tokens against 400K.
How many benchmarks do GPT-5 Mini and GPT-6 Luna share?
34 benchmarks have published results for both models. GPT-5 Mini has 60 scored results on Noometry and GPT-6 Luna has 42.