Model comparison
Gemma 3 4B vs Llama 4 Maverick
Llama 4 Maverick is the stronger model overall, scoring 30.6 to 27.9 on the Noometry Index. Gemma 3 4B costs 6.1× less per token, which makes it the better buy when Llama 4 Maverick's lead doesn't matter for your workload.
Last verified . 21 shared benchmarks.
Google
MetaSummary
- They share 21 benchmarks with published results for both. Gemma 3 4B scores higher in 5 categories and Llama 4 Maverick in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Llama 4 Maverick leads 33.3 to 11.8.
- The biggest single-benchmark swing is GPQA Diamond: 23.2% for Gemma 3 4B and 67% for Llama 4 Maverick.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.19 / $0.65 for Llama 4 Maverick.
- Gemma 3 4B accepts more context: 131K tokens versus 128K.
Side by side
| Gemma 3 4B | Llama 4 Maverick | |
|---|---|---|
| Provider | Meta | |
| Noometry Index | 27.9 | 30.6 |
| Released | 2025-03-12 | 2025-04-05 |
| Weights | Open | Open |
| Context window | 131K | 128K |
| Max output | 4K | 4K |
| Input $ / M tokens | $0.04 | $0.19 |
| Output $ / M tokens | $0.08 | $0.65 |
| Results tracked | 22 | 54 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 3 4B leads
Gemma 3 4B: 35.3 (#214), Llama 4 Maverick: 26.3 (#318)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| LMArena Coding | 1230 | 1302 |
| SWE-bench Verified (bash only) | — | 21% |
| Aider Polyglot | — | 15.6% |
| SciCode | — | 33.1% |
| WeirdML | — | 24.5% |
| BigCodeBench Instruct | — | 49.7% |
| BigCodeBench Complete | — | 61.4% |
| ALE-Bench | — | 172.97 |
Agentic & Tool Use Llama 4 Maverick leads
Gemma 3 4B: 20.7 (#134), Llama 4 Maverick: 27.9 (#86)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | 37.3% |
Reasoning Gemma 3 4B leads
Gemma 3 4B: 12.7 (#330), Llama 4 Maverick: 10.1 (#337)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| Kagi LLM Benchmark | 25.2% | 55.9% |
| LMArena Hard Prompts | 1253 | 1281 |
| DTBench | 50.9% | 61.9% |
| LMCA | 2.8% | 15.9% |
| Epoch Capabilities Index | 116.02 | 132.2 |
| ARC-AGI-2 | — | 0% |
| SimpleBench | — | 27.7% |
| NYT Connections (extended) | — | 8% |
| ARC-AGI-1 | — | 4.4% |
| CritPt | — | 0% |
| Chess Puzzles | 0% | — |
| EnigmaEval | — | 0.6% |
| ForecastBench | — | 57.5 |
Math Llama 4 Maverick leads
Gemma 3 4B: 15.8 (#286), Llama 4 Maverick: 24.6 (#256)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.5% | 20.6% |
| LMArena Math | 1239 | 1299 |
| Omni-MATH | — | 42.2% |
| MATH Level 5 | — | 73% |
| FrontierMath (Feb 2025 set) | — | 0.7% |
Knowledge Llama 4 Maverick leads
Gemma 3 4B: 11.8 (#294), Llama 4 Maverick: 33.3 (#196)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| GPQA Diamond | 23.2% | 67% |
| Vectara Hallucination Rate | 6.4% | 8.2% |
| LMArena Expert | 1223 | 1259 |
| Humanity's Last Exam | — | 5.7% |
| MMLU-Pro | — | 81% |
| Confabulations | — | 22.6% |
| GPQA (HELM) | — | 65% |
Multimodal Not comparable
Gemma 3 4B: —, Llama 4 Maverick: 30.9 (#100)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| LMArena Vision | — | 1141 |
| GeoBench | — | 52% |
| SpatialViz-Bench | — | 31.8% |
Multilingual Too close to call
Gemma 3 4B: 42.7 (#189), Llama 4 Maverick: 42.4 (#190)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| LMArena Non-English | 1273 | 1269 |
| LMArena German | 1281 | 1291 |
| LMArena Russian | 1294 | 1286 |
| LMArena Chinese | — | 1277 |
| LMArena French | — | 1258 |
| LMArena Japanese | — | 1207 |
| LMArena Korean | — | 1203 |
| LMArena Spanish | — | 1293 |
Instruction Following Llama 4 Maverick leads
Gemma 3 4B: 65.4 (#219), Llama 4 Maverick: 71.9 (#140)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| LMArena Instruction Following | 1239 | 1267 |
| IFEval | — | 90.8% |
Long Context Gemma 3 4B leads
Gemma 3 4B: 38.6 (#189), Llama 4 Maverick: 31.4 (#274)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| LMArena Longer Query | 1274 | 1280 |
| Fiction.LiveBench | — | 46.2% |
Writing & Preference Gemma 3 4B leads
Gemma 3 4B: 42.2 (#234), Llama 4 Maverick: 39.0 (#247)
| Benchmark | Gemma 3 4B | Llama 4 Maverick |
|---|---|---|
| LMArena Text | 1291 | 1287 |
| LMArena Creative Writing | 1270 | 1266 |
| EQ-Bench Creative Writing | 1068 | 860 |
| LMArena Multi-Turn | 1255 | 1289 |
| Short-Story Creative Writing | — | 62% |
| WildBench | — | 80% |
Frequently asked questions
Is Gemma 3 4B better than Llama 4 Maverick?
Llama 4 Maverick is the stronger model overall, scoring 30.6 to 27.9 on the Noometry Index. Gemma 3 4B costs 6.1× less per token, which makes it the better buy when Llama 4 Maverick's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 4B or Llama 4 Maverick?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; Llama 4 Maverick lists at $0.19 and $0.65.
Is Gemma 3 4B or Llama 4 Maverick better for coding?
Gemma 3 4B scores higher on coding benchmarks: 35.3 versus 26.3 in the Noometry coding category.
Which has the bigger context window?
Gemma 3 4B does, with 131K tokens against 128K.
How many benchmarks do Gemma 3 4B and Llama 4 Maverick share?
21 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and Llama 4 Maverick has 54.