Model comparison
Gemma 3 4B vs GPT-5.6 Luna
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 28.1 on the Noometry Index. Gemma 3 4B costs 9.0× less per token, which makes it the better buy when GPT-5.6 Luna's lead doesn't matter for your workload.
Last verified . 20 shared benchmarks.
GoogleSummary
- They share 20 benchmarks with published results for both. Gemma 3 4B scores higher in 0 categories and GPT-5.6 Luna in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.7 to 16.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 7.5% for Gemma 3 4B and 98.3% for GPT-5.6 Luna.
- Gemma 3 4B is cheaper at $0.04 / $0.08 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 131K.
- Gemma 3 4B has downloadable open weights; the other is API-only.
Side by side
| Gemma 3 4B | GPT-5.6 Luna | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 28.1 | 54.6 |
| Released | 2025-03-12 | 2026-07-09 |
| Weights | Open | Proprietary |
| Context window | 131K | 1.05M |
| Max output | 4K | 128K |
| Input $ / M tokens | $0.04 | $0.20 |
| Output $ / M tokens | $0.08 | $1.20 |
| Results tracked | 22 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Luna leads
Gemma 3 4B: 35.9 (#215), GPT-5.6 Luna: 54.5 (#28)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| LMArena Coding | 1230 | 1466 |
| DeepSWE | — | 67.2% |
| FrontierCode | — | 39.8% |
| CursorBench | — | 35.9% |
| LMArena WebDev | — | 1519 |
| SciCode | — | 53.6% |
| WeirdML | — | 60.9% |
| ALE-Bench | — | 1,667 |
Agentic & Tool Use GPT-5.6 Luna leads
Gemma 3 4B: 20.9 (#142), GPT-5.6 Luna: 34.4 (#45)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| APEX-Agents | — | 43% |
| Berkeley Function Calling Leaderboard | 19.6% | — |
| BALROG | — | 45.6% |
| GDP.pdf | — | 22.7% |
| Vending-Bench 2 | — | 4,095 |
Reasoning GPT-5.6 Luna leads
Gemma 3 4B: 13.2 (#335), GPT-5.6 Luna: 47.6 (#43)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| Kagi LLM Benchmark | 25.2% | 49.1% |
| Chess Puzzles | 0% | 40% |
| LMArena Hard Prompts | 1253 | 1451 |
| DTBench | 50.9% | 89.1% |
| LMCA | 2.8% | 48.5% |
| Epoch Capabilities Index | 116.02 | 156.39 |
| ARC-AGI-2 | — | 59.5% |
| SimpleBench | — | 46.8% |
| NYT Connections (extended) | — | 69.4% |
| ARC-AGI-1 | — | 88% |
| CritPt | — | 20.6% |
| Mystery Game Puzzles | — | 21% |
| Surface Evolver Bench | — | 61.9% |
Math GPT-5.6 Luna leads
Gemma 3 4B: 16.8 (#292), GPT-5.6 Luna: 77.7 (#14)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.5% | 98.3% |
| LMArena Math | 1239 | 1458 |
| FrontierMath (Tiers 1-3) | — | 82.1% |
| FrontierMath Tier 4 | — | 61% |
| ProofBench | — | 60% |
Knowledge GPT-5.6 Luna leads
Gemma 3 4B: 11.8 (#299), GPT-5.6 Luna: 58.5 (#34)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| GPQA Diamond | 23.2% | 91.6% |
| LMArena Expert | 1223 | 1478 |
| SimpleQA Verified | — | 41% |
| Vectara Hallucination Rate | 6.4% | — |
Multimodal Not comparable
Gemma 3 4B: —, GPT-5.6 Luna: 42.7 (#28)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| LMArena Vision | — | 1258 |
| Blueprint-Bench 2 | — | 22.6% |
| Furniture Assembly | — | 42.5% |
| LMArena Document | — | 1457 |
Multilingual GPT-5.6 Luna leads
Gemma 3 4B: 42.5 (#194), GPT-5.6 Luna: 52.8 (#78)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| LMArena Non-English | 1273 | 1417 |
| LMArena German | 1281 | 1454 |
| LMArena Russian | 1294 | 1428 |
| LMArena Chinese | — | 1470 |
| LMArena French | — | 1456 |
| LMArena Japanese | — | 1411 |
| LMArena Korean | — | 1415 |
| LMArena Spanish | — | 1448 |
Instruction Following GPT-5.6 Luna leads
Gemma 3 4B: 65.2 (#225), GPT-5.6 Luna: 75.6 (#57)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| LMArena Instruction Following | 1239 | 1437 |
Long Context GPT-5.6 Luna leads
Gemma 3 4B: 38.7 (#194), GPT-5.6 Luna: 43.9 (#82)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| LMArena Longer Query | 1273 | 1436 |
Writing & Preference GPT-5.6 Luna leads
Gemma 3 4B: 42.0 (#239), GPT-5.6 Luna: 68.0 (#29)
| Benchmark | Gemma 3 4B | GPT-5.6 Luna |
|---|---|---|
| LMArena Text | 1291 | 1431 |
| LMArena Creative Writing | 1271 | 1396 |
| EQ-Bench Creative Writing | 1068 | 1829 |
| LMArena Multi-Turn | 1255 | 1434 |
| EQ-Bench 4 | — | 1156 |
Frequently asked questions
Is Gemma 3 4B better than GPT-5.6 Luna?
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 28.1 on the Noometry Index. Gemma 3 4B costs 9.0× 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, Gemma 3 4B or GPT-5.6 Luna?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; GPT-5.6 Luna lists at $0.20 and $1.20.
Is Gemma 3 4B or GPT-5.6 Luna better for coding?
GPT-5.6 Luna scores higher on coding benchmarks: 54.5 versus 35.9 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Luna does, with 1.05M tokens against 131K.
How many benchmarks do Gemma 3 4B and GPT-5.6 Luna share?
20 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and GPT-5.6 Luna has 52.