Gemma 3 1B (IT): benchmark results
Google's Gemma 3 1B instruct model, fine-tuned from the Gemma 3 1B pretrained checkpoint under the Gemma license, released March 2025; the linked card provides no further details here. Provider: Google. Released 2025-03-12. Access: Open.
Aggregate Rating 1286 ± 8, rank #1504 of 1656 rated models, from 717 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LA Leaderboard - ClinTreatES | 63.7 | Semantic Answer Similarity (x100) | 86.6 |
| LA Leaderboard - QNLI Basque | 54.2 | Accuracy (%) | 78.7 |
| CVGEvalKit | 79.03 | ROUGE-1 | 78.6 |
| EuroEval Catalan Summarization - Dacsa CA | 35.53 | chrF3++ (%) | 74.4 |
| LA Leaderboard - ClinDiagnosES | 59.02 | Semantic Answer Similarity (x100) | 73.1 |
| LA Leaderboard - XNLI Basque | 37.36 | Accuracy (%) | 71.3 |
| LA Leaderboard - BEC2016 Basque | 41.86 | F1 (%) | 69.1 |
| LA Leaderboard - ParafrasesGL | 52.72 | Accuracy (%) | 69.1 |
| LA Leaderboard - NoticIA | 17.52 | ROUGE-1 (%) | 66.2 |
| Kurakurai French LLM - MBPP | 43.86 | pass@1 (%) | 65.9 |
| EuroEval Italian Summarization - Ilpost SUM | 37.08 | chrF3++ (%) | 63.3 |
| EuroEval Serbian Summarization - LR SUM SR | 27.92 | chrF3++ (%) | 61.2 |
Interactive version: theaggregate.ai/model?slug=gemma-3-1b-it · How It Works · Data refreshed daily, snapshot 2026-10-20.