Model comparison
Gemma 3 12B vs Qwen3.8 27B
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 32.1 on the Noometry Index. Gemma 3 12B costs 15× less per token, which makes it the better buy when Qwen3.8 27B's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
Google
Alibaba (Qwen)Summary
- They share 18 benchmarks with published results for both. Gemma 3 12B scores higher in 0 categories and Qwen3.8 27B in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.8 27B leads 41.0 to 15.7.
- The biggest single-benchmark swing is DTBench: 48.8% for Gemma 3 12B and 88% for Qwen3.8 27B.
- Gemma 3 12B is cheaper at $0.05 / $0.15 per million input/output tokens, against $0.99 / $1.49 for Qwen3.8 27B.
- Qwen3.8 27B accepts more context: 262K tokens versus 131K.
Side by side
| Gemma 3 12B | Qwen3.8 27B | |
|---|---|---|
| Provider | Alibaba (Qwen) | |
| Noometry Index | 32.1 | 46.0 |
| Released | 2025-03-12 | 2026-08-14 |
| Weights | Open | Open |
| Context window | 131K | 262K |
| Max output | 8K | 33K |
| Input $ / M tokens | $0.05 | $0.99 |
| Output $ / M tokens | $0.15 | $1.49 |
| Results tracked | 24 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 27B leads
Gemma 3 12B: 31.7 (#280), Qwen3.8 27B: 50.5 (#44)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| SciCode | 17.4% | 46.6% |
| LMArena Coding | 1281 | 1482 |
| LMArena WebDev | — | 1593 |
Agentic & Tool Use Qwen3.8 27B leads
Gemma 3 12B: 25.5 (#108), Qwen3.8 27B: 32.9 (#57)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| APEX-Agents | — | 47.5% |
| Berkeley Function Calling Leaderboard | 30.4% | — |
Reasoning Qwen3.8 27B leads
Gemma 3 12B: 15.7 (#313), Qwen3.8 27B: 41.0 (#54)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| CritPt | 0% | 5.4% |
| LMArena Hard Prompts | 1309 | 1460 |
| DTBench | 48.8% | 88% |
| LMCA | 4.5% | 41.4% |
| Epoch Capabilities Index | 123.5 | 149.38 |
| ARC-AGI-2 | — | 42.4% |
| NYT Connections (extended) | — | 54.5% |
| ARC-AGI-1 | — | 87.5% |
| Chess Puzzles | 0% | — |
| Surface Evolver Bench | — | 45% |
Math Qwen3.8 27B leads
Gemma 3 12B: 22.3 (#279), Qwen3.8 27B: 37.1 (#161)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| LMArena Math | 1307 | 1456 |
| OTIS Mock AIME 2024-2025 | 16.7% | — |
| ProofBench | — | 16% |
Knowledge Qwen3.8 27B leads
Gemma 3 12B: 26.5 (#257), Qwen3.8 27B: 41.6 (#109)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1248 | 1482 |
| GPQA Diamond | 39.5% | — |
| Vectara Hallucination Rate | 4.4% | — |
Multimodal Not comparable
Gemma 3 12B: —, Qwen3.8 27B: 41.3 (#37)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | — | 1271 |
| MindCube | 46.7% | — |
Multilingual Qwen3.8 27B leads
Gemma 3 12B: 45.7 (#165), Qwen3.8 27B: 53.7 (#60)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1318 | 1430 |
| LMArena German | 1370 | 1438 |
| LMArena Russian | 1335 | 1415 |
| LMArena Chinese | — | 1504 |
| LMArena French | — | 1465 |
| LMArena Japanese | — | 1384 |
| LMArena Korean | — | 1393 |
| LMArena Spanish | — | 1448 |
Instruction Following Qwen3.8 27B leads
Gemma 3 12B: 68.6 (#186), Qwen3.8 27B: 75.8 (#53)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1299 | 1439 |
Long Context Qwen3.8 27B leads
Gemma 3 12B: 40.0 (#162), Qwen3.8 27B: 44.3 (#70)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1317 | 1450 |
Writing & Preference Qwen3.8 27B leads
Gemma 3 12B: 47.5 (#209), Qwen3.8 27B: 65.8 (#43)
| Benchmark | Gemma 3 12B | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1334 | 1441 |
| LMArena Creative Writing | 1331 | 1384 |
| EQ-Bench Creative Writing | 1126 | 1671 |
| LMArena Multi-Turn | 1334 | 1441 |
Frequently asked questions
Is Gemma 3 12B better than Qwen3.8 27B?
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 32.1 on the Noometry Index. Gemma 3 12B costs 15× less per token, which makes it the better buy when Qwen3.8 27B's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 12B or Qwen3.8 27B?
Gemma 3 12B is cheaper. It lists at $0.05 per million input tokens and $0.15 per million output tokens; Qwen3.8 27B lists at $0.99 and $1.49.
Is Gemma 3 12B or Qwen3.8 27B better for coding?
Qwen3.8 27B scores higher on coding benchmarks: 50.5 versus 31.7 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 27B does, with 262K tokens against 131K.
How many benchmarks do Gemma 3 12B and Qwen3.8 27B share?
18 benchmarks have published results for both models. Gemma 3 12B has 24 scored results on Noometry and Qwen3.8 27B has 31.