Model comparison
Gemma 3 4B vs GPT-6 Luna
GPT-6 Luna is the stronger model overall, scoring 53.3 to 28.1 on the Noometry Index. Gemma 3 4B costs 4.0× less per token, which makes it the better buy when GPT-6 Luna's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
GoogleSummary
- They share 18 benchmarks with published results for both. Gemma 3 4B scores higher in 0 categories and GPT-6 Luna in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Luna leads 76.1 to 16.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 7.5% for Gemma 3 4B and 98.9% for GPT-6 Luna.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.10 / $0.50 for GPT-6 Luna.
- GPT-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-6 Luna | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 28.1 | 53.3 |
| Released | 2025-03-12 | 2026-09-22 |
| Weights | Open | Proprietary |
| Context window | 131K | 1.05M |
| Max output | 4K | 128K |
| Input $ / M tokens | $0.04 | $0.10 |
| Output $ / M tokens | $0.08 | $0.50 |
| Results tracked | 22 | 42 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Luna leads
Gemma 3 4B: 35.9 (#215), GPT-6 Luna: 55.5 (#25)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| LMArena Coding | 1230 | 1439 |
| DeepSWE | — | 66.6% |
| FrontierCode | — | 42.4% |
| LMArena WebDev | — | 1581 |
| SciCode | — | 54.6% |
| ALE-Bench | — | 1,577 |
Agentic & Tool Use GPT-6 Luna leads
Gemma 3 4B: 20.9 (#142), GPT-6 Luna: 33.3 (#54)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| APEX-Agents | — | 44.3% |
| Berkeley Function Calling Leaderboard | 19.6% | — |
| GDP.pdf | — | 23% |
Reasoning GPT-6 Luna leads
Gemma 3 4B: 13.2 (#335), GPT-6 Luna: 48.2 (#41)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| Chess Puzzles | 0% | 31% |
| LMArena Hard Prompts | 1253 | 1411 |
| DTBench | 50.9% | 90.1% |
| LMCA | 2.8% | 44.5% |
| Epoch Capabilities Index | 116.02 | 156.28 |
| ARC-AGI-2 | — | 59.3% |
| Kagi LLM Benchmark | 25.2% | — |
| NYT Connections (extended) | — | 68.7% |
| ARC-AGI-1 | — | 86.7% |
| CritPt | — | 19.4% |
| Mystery Game Puzzles | — | 7% |
Math GPT-6 Luna leads
Gemma 3 4B: 16.8 (#292), GPT-6 Luna: 76.1 (#15)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.5% | 98.9% |
| LMArena Math | 1239 | 1416 |
| FrontierMath (Tiers 1-3) | — | 78.9% |
| FrontierMath Tier 4 | — | 56.1% |
| ProofBench | — | 64% |
Knowledge GPT-6 Luna leads
Gemma 3 4B: 11.8 (#299), GPT-6 Luna: 57.0 (#41)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| GPQA Diamond | 23.2% | 90.5% |
| LMArena Expert | 1223 | 1444 |
| SimpleQA Verified | — | 41.4% |
| Vectara Hallucination Rate | 6.4% | — |
Multimodal Not comparable
Gemma 3 4B: —, GPT-6 Luna: 42.4 (#30)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| LMArena Vision | — | 1217 |
| Blueprint-Bench 2 | — | 31.2% |
| Furniture Assembly | — | 44.2% |
Multilingual GPT-6 Luna leads
Gemma 3 4B: 42.5 (#194), GPT-6 Luna: 50.5 (#117)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| LMArena Non-English | 1273 | 1386 |
| LMArena German | 1281 | 1369 |
| LMArena Russian | 1294 | 1394 |
| LMArena Chinese | — | 1433 |
| LMArena French | — | 1420 |
| LMArena Japanese | — | 1369 |
| LMArena Korean | — | 1360 |
| LMArena Spanish | — | 1393 |
Instruction Following GPT-6 Luna leads
Gemma 3 4B: 65.2 (#225), GPT-6 Luna: 74.3 (#99)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| LMArena Instruction Following | 1239 | 1409 |
Long Context GPT-6 Luna leads
Gemma 3 4B: 38.7 (#194), GPT-6 Luna: 43.0 (#111)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| LMArena Longer Query | 1273 | 1409 |
Writing & Preference GPT-6 Luna leads
Gemma 3 4B: 42.0 (#239), GPT-6 Luna: 58.3 (#119)
| Benchmark | Gemma 3 4B | GPT-6 Luna |
|---|---|---|
| LMArena Text | 1291 | 1391 |
| LMArena Creative Writing | 1271 | 1363 |
| LMArena Multi-Turn | 1255 | 1396 |
| EQ-Bench Creative Writing | 1068 | — |
Frequently asked questions
Is Gemma 3 4B better than GPT-6 Luna?
GPT-6 Luna is the stronger model overall, scoring 53.3 to 28.1 on the Noometry Index. Gemma 3 4B costs 4.0× less per token, which makes it the better buy when GPT-6 Luna's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 4B or GPT-6 Luna?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; GPT-6 Luna lists at $0.10 and $0.50.
Is Gemma 3 4B or GPT-6 Luna better for coding?
GPT-6 Luna scores higher on coding benchmarks: 55.5 versus 35.9 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Luna does, with 1.05M tokens against 131K.
How many benchmarks do Gemma 3 4B and GPT-6 Luna share?
18 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and GPT-6 Luna has 42.