Model comparison
Gemma 3 12B vs Step 3
Step 3 is the stronger model overall, scoring 40.5 to 32.1 on the Noometry Index.
Last verified . 12 shared benchmarks.
Google
StepFunSummary
- They share 12 benchmarks with published results for both. Gemma 3 12B scores higher in 0 categories and Step 3 in 8 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where Step 3 leads 37.6 to 22.3.
Side by side
| Gemma 3 12B | Step 3 | |
|---|---|---|
| Provider | StepFun | |
| Noometry Index | 32.1 | 40.5 |
| Released | 2025-03-12 | — |
| Weights | Open | Open |
| Context window | 131K | — |
| Max output | 8K | — |
| Input $ / M tokens | $0.05 | — |
| Output $ / M tokens | $0.15 | — |
| Results tracked | 24 | 17 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 3 leads
Gemma 3 12B: 31.7 (#280), Step 3: 40.1 (#147)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Coding | 1281 | 1367 |
| SciCode | 17.4% | — |
Agentic & Tool Use Not comparable
Gemma 3 12B: 25.5 (#108), Step 3: —
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 30.4% | — |
Reasoning Step 3 leads
Gemma 3 12B: 15.7 (#313), Step 3: 28.4 (#105)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Hard Prompts | 1309 | 1355 |
| Kagi LLM Benchmark | — | 62.3% |
| CritPt | 0% | — |
| Chess Puzzles | 0% | — |
| DTBench | 48.8% | — |
| LMCA | 4.5% | — |
| Epoch Capabilities Index | 123.5 | — |
Math Step 3 leads
Gemma 3 12B: 22.3 (#279), Step 3: 37.6 (#148)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Math | 1307 | 1366 |
| OTIS Mock AIME 2024-2025 | 16.7% | — |
Knowledge Step 3 leads
Gemma 3 12B: 26.5 (#257), Step 3: 36.8 (#164)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Expert | 1248 | 1333 |
| GPQA Diamond | 39.5% | — |
| Vectara Hallucination Rate | 4.4% | — |
Multimodal Not comparable
Gemma 3 12B: —, Step 3: 35.5 (#86)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Vision | — | 1177 |
| MindCube | 46.7% | — |
Multilingual Too close to call
Gemma 3 12B: 45.7 (#165), Step 3: 46.3 (#159)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Non-English | 1318 | 1327 |
| LMArena German | 1370 | 1371 |
| LMArena Russian | 1335 | 1331 |
| LMArena Chinese | — | 1397 |
| LMArena Korean | — | 1269 |
| LMArena Spanish | — | 1371 |
Instruction Following Step 3 leads
Gemma 3 12B: 68.6 (#186), Step 3: 70.4 (#164)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1299 | 1332 |
Long Context Too close to call
Gemma 3 12B: 40.0 (#162), Step 3: 40.3 (#157)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Longer Query | 1317 | 1326 |
Writing & Preference Step 3 leads
Gemma 3 12B: 47.5 (#209), Step 3: 54.3 (#151)
| Benchmark | Gemma 3 12B | Step 3 |
|---|---|---|
| LMArena Text | 1334 | 1350 |
| LMArena Creative Writing | 1331 | 1321 |
| LMArena Multi-Turn | 1334 | 1341 |
| EQ-Bench Creative Writing | 1126 | — |
Frequently asked questions
Is Gemma 3 12B better than Step 3?
Step 3 is the stronger model overall, scoring 40.5 to 32.1 on the Noometry Index.
Is Gemma 3 12B or Step 3 better for coding?
Step 3 scores higher on coding benchmarks: 40.1 versus 31.7 in the Noometry coding category.
How many benchmarks do Gemma 3 12B and Step 3 share?
12 benchmarks have published results for both models. Gemma 3 12B has 24 scored results on Noometry and Step 3 has 17.