Nemotron 3.5 Bolt E4B: benchmark results
NVIDIA's Nemotron 3.5 Bolt E4B (Oct 23, 2026), post-trained from Gemma 4 E4B (8.0B total, 4B effective parameters). Provider: NVIDIA. Released 2026-10-23. Access: Open.
Aggregate Rating 1596 ± 49, rank #962 of 10516 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SEAL - MultiChallenge (Small Language Models) | 56.03 | Score | 100 |
| Docbench - Narration | 83 | Native task quality score (%) | 53.6 |
| Docbench - Writing | 85 | Native task quality score (%) | 48.2 |
| Docbench - Review | 50 | Native task quality score (%) | 44.6 |
| SEA-HELM | 60.56 | Mean Score (%) | 44.6 |
| Docbench - Study | 68 | Native task quality score (%) | 42.9 |
| Docbench - Overall | 69 | Native task quality score (%) | 39.3 |
| Docbench - Slides | 82 | Native task quality score (%) | 37.5 |
| Docbench - Diagrams | 66 | Native task quality score (%) | 25 |
Interactive version: theaggregate.ai/model?slug=nemotron-3-5-bolt-e4b · How It Works · Data refreshed daily, snapshot 2026-11-21.