Gemma 4 12B (Reasoning) is a Google Gemma open-model profile, aimed at developers who want deployable Google-developed models rather than hosted Gemini-only access. The benchmark data is especially useful for comparing its efficiency and capability against similarly sized open and open-weight alternatives.
Introducing GemmaRank #225 across 600
Rank #164 across 266
Rank #34 across 379
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Higher bars mean stronger relative placement.
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 21.8 | #225 |
| Artificial Analysis Coding Index | coding | 31.0 | #164 |
| GPQA | reasoning | 75.3% | #213 |
| Humanity's Last Exam | reasoning | 14.8% | #154 |
| SciCode |
| coding, reasoning |
| 38.2% |
| #192 |
| Output Speed | speed | 122.2 tok/s | #65 |
| Time to First Token | speed | 1.40s | #93 |
| Blended Price | cost | $0.150/M | #34 |
| Input Price | cost | $0.100/M | #38 |
| Output Price | cost | $0.300/M | #40 |
| Value Index | cost, overall | 145.3 | #20 |