Google: Gemma 4 31B wins on more metrics (6 of 9), but the right pick depends on what you optimize for — see the breakdown below.
| Metric | Google: Gemma 4 31B | OpenAI: GPT-5.1-Codex-Mini |
|---|---|---|
| Intelligence Index | 39.2✓ | 38.6 |
| Coding Index | 38.7✓ | 36.4 |
| GPQA Diamond | 86%✓ | 81% |
| Design Arena Elo | — | 1162✓ |
| Speed (tokens/sec) | 65 | 121✓ |
| Latency | 266ms✓ | 1.7s |
| Input price /M | $0.120✓ | $0.250 |
| Output price /M | $0.350✓ | $2.00 |
| Context window | 262K | 400K✓ |
| Capabilities | ReasoningToolsJSONVision | ReasoningToolsJSONVision |