Gemma 4 26B A4B (Non-reasoning): benchmark results
Provider: Google. Released 2026-04-02. Access: Open.
Aggregate Rating 1541 ± 20, rank #933 of 2201 rated models, from 39 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ObGynLongBench - Evidence-Only | 78.2 | Accuracy (%; supporting evidence only) | 87.5 |
| ObGynLongBench - History-Level EHR - L1 Single Evidence | 65 | Accuracy (%) | 81.2 |
| ObGynLongBench - Visit-Level EHR | 60.6 | Accuracy (%; same-day visit records) | 81.2 |
| VAMPS | 85.3 | Accuracy (%) on the English versions of the 218 Konkour-seed | 76.5 |
| VAMPS (Analytical) - Persian | 95.4 | Accuracy (%) on the original Persian versions of the 218 Kon | 70.6 |
| AA Omniscience - Software Engineering (SWE) - PHP | 28 | Accuracy (%) | 69.1 |
| ObGynLongBench - History-Level EHR | 60.7 | Accuracy (%; full pre-decision EHR history) | 68.8 |
| ObGynLongBench - History-Level EHR - L2 Multi-Source Integration | 61.4 | Accuracy (%) | 68.8 |
| AA Terminal-Bench Hard | 25 | Accuracy (%) | 66 |
| VAMPS - Persian | 81.2 | Accuracy (%) on the original Persian versions of the 218 Kon | 64.7 |
| AA Omniscience - Software Engineering (SWE) - Rust | 50 | Accuracy (%) | 59 |
| AA Omniscience - Software Engineering (SWE) - Go | 18 | Accuracy (%) | 56.5 |
Interactive version: theaggregate.ai/model?slug=gemma-4-26b-a4b-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-12.