Gemma 4 31B (IT) (Non-reasoning): benchmark results
Provider: Google. Released 2026-04-02. Access: Open.
Aggregate Rating 1633 ± 12, rank #590 of 2170 rated models, from 81 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CDR-Bench - Order-Agnostic Recipes | 25.55 | Recipe success RS@3 (%) on the order-agnostic mapper-recipe | 100 |
| KG-to-Text 2026 - Coverage WebNLG20 GLM Judge | 1 | Mean score (0–1) | 100 |
| KG-to-Text 2026 - Faithfulness DART DeepSeek Judge | 0.98 | Mean score (0–1) | 100 |
| KG-to-Text 2026 - Faithfulness DART GLM Judge | 0.99 | Mean score (0–1) | 100 |
| KG-to-Text 2026 - Faithfulness WebNLG17 DeepSeek Judge | 0.97 | Mean score (0–1) | 100 |
| KG-to-Text 2026 - Faithfulness WebNLG17 GLM Judge | 0.99 | Mean score (0–1) | 100 |
| KG-to-Text 2026 - Faithfulness WebNLG20 DeepSeek Judge | 0.97 | Mean score (0–1) | 100 |
| KG-to-Text 2026 - Faithfulness WebNLG20 GLM Judge | 0.99 | Mean score (0–1) | 100 |
| KG-to-Text 2026 - WebNLG20 BLEU | 54.85 | BLEU | 100 |
| KG-to-Text 2026 - WebNLG20 ChrF++ | 70.65 | ChrF++ | 100 |
| KG-to-Text 2026 - WebNLG20 METEOR | 0.76 | METEOR | 100 |
| KG-to-Text 2026 - WebNLG20 PARENT F1 | 0.66 | PARENT F1 | 100 |
Interactive version: theaggregate.ai/model?slug=gemma-4-31b-it-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-05.