Model comparison
Gemma 2 9B vs Gemma 3 4B
Gemma 3 4B is the stronger model overall, scoring 28.1 to 25.9 on the Noometry Index.
Last verified . 16 shared benchmarks.
GoogleSummary
- They share 16 benchmarks with published results for both. Gemma 2 9B scores higher in 1 category and Gemma 3 4B in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Gemma 3 4B leads 42.0 to 32.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 0.6% for Gemma 2 9B and 7.5% for Gemma 3 4B.
Side by side
| Gemma 2 9B | Gemma 3 4B | |
|---|---|---|
| Provider | ||
| Noometry Index | 25.9 | 28.1 |
| Released | 2024-06-24 | 2025-03-12 |
| Weights | Open | Open |
| Context window | — | 131K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.04 |
| Output $ / M tokens | — | $0.08 |
| Results tracked | 35 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 3 4B leads
Gemma 2 9B: 29.4 (#304), Gemma 3 4B: 35.9 (#215)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| LMArena Coding | 1173 | 1230 |
| BigCodeBench Instruct | 34.7% | — |
| LiveBench Coding | 22.5% | — |
| BigCodeBench Complete | 40.6% | — |
Agentic & Tool Use Not comparable
Gemma 2 9B: —, Gemma 3 4B: 20.9 (#142)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 19.6% |
Reasoning Gemma 2 9B leads
Gemma 2 9B: 15.9 (#309), Gemma 3 4B: 13.2 (#335)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| LMArena Hard Prompts | 1171 | 1253 |
| Epoch Capabilities Index | 119.83 | 116.02 |
| Kagi LLM Benchmark | — | 25.2% |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | 15.2% | — |
| DTBench | — | 50.9% |
| LiveBench Data Analysis | 36.4% | — |
| LMCA | — | 2.8% |
| LiveBench | 28.7% | — |
| PIQA | 83.7% | — |
Math Gemma 3 4B leads
Gemma 2 9B: 9.9 (#318), Gemma 3 4B: 16.8 (#292)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 0.6% | 7.5% |
| LMArena Math | 1183 | 1239 |
| LiveBench Math | 19.8% | — |
| MATH Level 5 | 21% | — |
| GSM8K | 84.9% | — |
Knowledge Gemma 3 4B leads
Gemma 2 9B: 9.7 (#305), Gemma 3 4B: 11.8 (#299)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| GPQA Diamond | 27.5% | 23.2% |
| LMArena Expert | 1147 | 1223 |
| Vectara Hallucination Rate | — | 6.4% |
| BoolQ | 85.7% | — |
| MMLU | 72.1% | — |
Multilingual Gemma 3 4B leads
Gemma 2 9B: 36.6 (#238), Gemma 3 4B: 42.5 (#194)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| LMArena Non-English | 1188 | 1273 |
| LMArena German | 1186 | 1281 |
| LMArena Russian | 1200 | 1294 |
| LMArena Chinese | 1185 | — |
| LMArena French | 1190 | — |
| LMArena Japanese | 1144 | — |
| LMArena Korean | 1137 | — |
| LMArena Spanish | 1200 | — |
Instruction Following Gemma 3 4B leads
Gemma 2 9B: 57.6 (#269), Gemma 3 4B: 65.2 (#225)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| LMArena Instruction Following | 1178 | 1239 |
| LiveBench Instruction Following | 52.6% | — |
Long Context Gemma 3 4B leads
Gemma 2 9B: 36.3 (#233), Gemma 3 4B: 38.7 (#194)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| LMArena Longer Query | 1197 | 1273 |
Writing & Preference Gemma 3 4B leads
Gemma 2 9B: 32.1 (#281), Gemma 3 4B: 42.0 (#239)
| Benchmark | Gemma 2 9B | Gemma 3 4B |
|---|---|---|
| LMArena Text | 1207 | 1291 |
| LMArena Creative Writing | 1206 | 1271 |
| EQ-Bench Creative Writing | 841 | 1068 |
| LMArena Multi-Turn | 1193 | 1255 |
| LiveBench Language | 25.5% | — |
Frequently asked questions
Is Gemma 2 9B better than Gemma 3 4B?
Gemma 3 4B is the stronger model overall, scoring 28.1 to 25.9 on the Noometry Index.
Is Gemma 2 9B or Gemma 3 4B better for coding?
Gemma 3 4B scores higher on coding benchmarks: 35.9 versus 29.4 in the Noometry coding category.
How many benchmarks do Gemma 2 9B and Gemma 3 4B share?
16 benchmarks have published results for both models. Gemma 2 9B has 35 scored results on Noometry and Gemma 3 4B has 22.