Google: Gemma 4 31B wins on more metrics (5 of 7), but the right pick depends on what you optimize for — see the breakdown below.
| Metric | Google: Gemma 4 31B | Qwen: Qwen3 Max Thinking |
|---|---|---|
| Intelligence Index | 39.2 | 39.8✓ |
| Coding Index | 38.7✓ | 30.5 |
| GPQA Diamond | 86% | 86%✓ |
| Design Arena Elo | — | — |
| Speed (tokens/sec) | 65✓ | 38 |
| Latency | 266ms✓ | 1.3s |
| Input price /M | $0.120✓ | $0.780 |
| Output price /M | $0.350✓ | $3.90 |
| Context window | 262K | 262K |
| Capabilities | ReasoningToolsJSONVision | ReasoningToolsJSON |