Gemma 4 31B: benchmark results
Google's flagship dense Gemma 4 31B open model. Provider: Google. Released 2026-04-02. Access: Open.
Aggregate Rating 1697 ± 5, rank #200 of 1656 rated models, from 1037 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AMIGO (Threshold 0.3) | 25.2 | Verified Correct accuracy in percent on AMIGO's dress galler | 100 |
| AMIGO (Threshold 0.4) | 27.9 | Verified Correct accuracy in percent on AMIGO's dress galler | 100 |
| AMIGO (Threshold 0.5) | 26.6 | Verified Correct accuracy in percent on AMIGO's dress galler | 100 |
| AMIGO (Threshold 0.6) | 18.6 | Verified Correct accuracy in percent on AMIGO's dress galler | 100 |
| BARISTA - Hand-Object Interaction | 75.8 | Interaction F1 (%, times 100) for recognizing which hand int | 100 |
| BARISTA - Relation Extraction | 75.1 | Micro F1 (%, times 100) of extracted subject, target, relati | 100 |
| BenSyc | 61.7 | Five-class conversational alignment classification macro-F1 | 100 |
| BloomBench - Arabic | 87.6 | Micro accuracy (%) on all 7,747 Modern Standard Arabic Bloom | 100 |
| BloomBench - Arabic - Analyze | 91 | Accuracy (%) on the 1431 Arabic BloomBench items at the Bloo | 100 |
| BloomBench - Arabic - Create | 82 | Accuracy (%) on the 685 Arabic BloomBench items at the Bloom | 100 |
| BloomBench - Arabic - Evaluate | 93 | Accuracy (%) on the 592 Arabic BloomBench items at the Bloom | 100 |
| BloomBench - Arabic - Remember | 85 | Accuracy (%) on the 2948 Arabic BloomBench items at the Bloo | 100 |
Interactive version: theaggregate.ai/model?slug=gemma-4-31b · How It Works · Data refreshed daily, snapshot 2026-10-20.