Model comparison
Gemma 2 27B vs Gemma 3 27B
Gemma 3 27B is the stronger model overall, scoring 30.8 to 29.4 on the Noometry Index.
Last verified . 31 shared benchmarks.
GoogleSummary
- They share 31 benchmarks with published results for both. Gemma 2 27B scores higher in 2 categories and Gemma 3 27B in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 3 27B leads 25.9 to 10.7.
- The biggest single-benchmark swing is MATH Level 5: 27.9% for Gemma 2 27B and 74% for Gemma 3 27B.
- Gemma 3 27B is cheaper at $0.08 / $0.16 per million input/output tokens, against $0.65 / $0.65 for Gemma 2 27B.
- Gemma 3 27B accepts more context: 131K tokens versus 8K.
Side by side
| Gemma 2 27B | Gemma 3 27B | |
|---|---|---|
| Provider | ||
| Noometry Index | 29.4 | 30.8 |
| Released | 2024-06-24 | 2025-03-11 |
| Weights | Open | Open |
| Context window | 8K | 131K |
| Max output | 2K | 8K |
| Input $ / M tokens | $0.65 | $0.08 |
| Output $ / M tokens | $0.65 | $0.16 |
| Results tracked | 34 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 2 27B leads
Gemma 2 27B: 34.1 (#246), Gemma 3 27B: 22.5 (#334)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| LiveBench Coding | 36% | 39.9% |
| LMArena Coding | 1211 | 1322 |
| Aider Polyglot | — | 4.9% |
| SciCode | — | 21.2% |
| BigCodeBench Instruct | 42.8% | — |
| BigCodeBench Complete | 52.5% | — |
Agentic & Tool Use Not comparable
Gemma 2 27B: —, Gemma 3 27B: 25.1 (#110)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 29.5% |
Reasoning Gemma 3 27B leads
Gemma 2 27B: 15.3 (#315), Gemma 3 27B: 16.7 (#301)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| LiveBench Reasoning | 28.1% | 43.8% |
| LMArena Hard Prompts | 1198 | 1340 |
| DTBench | 48% | 52.5% |
| LiveBench Data Analysis | 47.9% | 51.5% |
| LMCA | 7.1% | 12.3% |
| Epoch Capabilities Index | 122.08 | 130.04 |
| LiveBench | 38.2% | 50% |
| Kagi LLM Benchmark | — | 40.4% |
| CritPt | — | 0% |
| Chess Puzzles | — | 0% |
Math Gemma 3 27B leads
Gemma 2 27B: 10.7 (#311), Gemma 3 27B: 25.9 (#265)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.4% | 22.5% |
| LiveBench Math | 26.5% | 55.4% |
| LMArena Math | 1212 | 1312 |
| MATH Level 5 | 27.9% | 74% |
Knowledge Gemma 3 27B leads
Gemma 2 27B: 19.0 (#280), Gemma 3 27B: 25.5 (#261)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| GPQA Diamond | 36.5% | 47.7% |
| Confabulations | 27.1% | 40.3% |
| LMArena Expert | 1172 | 1304 |
| Vectara Hallucination Rate | — | 7.4% |
| MMLU | 75.7% | — |
Multimodal Not comparable
Gemma 2 27B: —, Gemma 3 27B: 32.6 (#100)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| LMArena Vision | — | 1164 |
| GeoBench | — | 52% |
Multilingual Gemma 3 27B leads
Gemma 2 27B: 38.6 (#226), Gemma 3 27B: 46.9 (#155)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| LMArena Non-English | 1217 | 1334 |
| LMArena Chinese | 1221 | 1346 |
| LMArena French | 1247 | 1368 |
| LMArena German | 1209 | 1362 |
| LMArena Japanese | 1175 | 1287 |
| LMArena Korean | 1174 | 1308 |
| LMArena Russian | 1234 | 1349 |
| LMArena Spanish | 1228 | 1349 |
Instruction Following Gemma 3 27B leads
Gemma 2 27B: 60.5 (#249), Gemma 3 27B: 70.6 (#160)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| LiveBench Instruction Following | 58.1% | 74.9% |
| LMArena Instruction Following | 1206 | 1321 |
Long Context Gemma 2 27B leads
Gemma 2 27B: 37.3 (#218), Gemma 3 27B: 27.6 (#293)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| LMArena Longer Query | 1231 | 1333 |
| Fiction.LiveBench | — | 33.3% |
Writing & Preference Gemma 3 27B leads
Gemma 2 27B: 44.2 (#225), Gemma 3 27B: 52.5 (#168)
| Benchmark | Gemma 2 27B | Gemma 3 27B |
|---|---|---|
| LMArena Text | 1231 | 1358 |
| LMArena Creative Writing | 1241 | 1346 |
| LMArena Multi-Turn | 1224 | 1345 |
| LiveBench Language | 32.6% | 34.6% |
| Short-Story Creative Writing | — | 79.9% |
| EQ-Bench Creative Writing | — | 1266 |
Frequently asked questions
Is Gemma 2 27B better than Gemma 3 27B?
Gemma 3 27B is the stronger model overall, scoring 30.8 to 29.4 on the Noometry Index.
Which is cheaper, Gemma 2 27B or Gemma 3 27B?
Gemma 3 27B is cheaper. It lists at $0.08 per million input tokens and $0.16 per million output tokens; Gemma 2 27B lists at $0.65 and $0.65.
Is Gemma 2 27B or Gemma 3 27B better for coding?
Gemma 2 27B scores higher on coding benchmarks: 34.1 versus 22.5 in the Noometry coding category.
Which has the bigger context window?
Gemma 3 27B does, with 131K tokens against 8K.
How many benchmarks do Gemma 2 27B and Gemma 3 27B share?
31 benchmarks have published results for both models. Gemma 2 27B has 34 scored results on Noometry and Gemma 3 27B has 43.