Gemma 4 26B A4B: benchmark results
Google's Apache-2.0 Gemma 4 MoE (April 2026) with 26B total and ~4B active parameters, multimodal with context up to 256K. Provider: Google. Released 2026-04-02. Access: Open.
Aggregate Rating 1630 ± 7, rank #358 of 1656 rated models, from 705 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| NVIDIA Nemotron 3.5 Lightning Card - IFBench (loose) | 77.25 | Score (%) | 100 |
| NVIDIA Nemotron 3.5 Lightning Card - SciCode | 40.28 | Score (%) | 100 |
| NVIDIA Nemotron 3.5 Lightning Card - τ³-bench (Banking) | 14.02 | Score (%) | 100 |
| Xing4.0 Card - AA.LCR | 66 | Score (%) | 100 |
| Xing4.0 Card - IFBench | 72.67 | Score (%) | 100 |
| Medmarks - SCTpublic | 75.34 | Score (%) | 97.7 |
| aiXamine - Fairness & Bias | 77.5 | Service score (0-100; normalized pass rate) | 97.6 |
| Medmarks - MedXpertQA Reasoning | 53.13 | Score (%) | 96.6 |
| Medmarks - MedXpertQA Understanding | 44.2 | Score (%) | 95.5 |
| Kaggle Open Benchmarks - LiveCodeBench V6 | 90.7 | Pass Rate (%) | 95.3 |
| Medmarks - MedHallu Hard | 61.76 | Score (%) | 94.3 |
| Medmarks - MedQA | 93.53 | Score (%) | 94.3 |
Interactive version: theaggregate.ai/model?slug=gemma-4-26b-a4b · How It Works · Data refreshed daily, snapshot 2026-10-20.