Ornith-1.5-9B: benchmark results
Ornith's Ornith-1.5-9B, a 9B dense model refined through a self-improving reinforcement-learning loop that jointly optimizes task generation, scaffolding and rollouts, sized for single-GPU or edge deployment. Provider: Ornith. Released 2026-08-19. Access: Open.
Aggregate Rating 1533 ± 24, rank #959 of 2170 rated models, from 34 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Nejumi 4 - GLP - Logical Reasoning | 94.5 | Score (%) | 92.6 |
| Nejumi 4 - ALT - Toxicity | 83.29 | Score (%) | 72.1 |
| Nejumi 4 - GLP - Expression | 97.5 | Score (%) | 66.9 |
| Nejumi 4 - ALT - Truthfulness | 78.81 | Score (%) | 66.2 |
| Nejumi 4 - ALT - Bias | 88.1 | Score (%) | 58.8 |
| Nejumi 4 - MT-Bench (Japanese) | 95.38 | Score (%) | 55.5 |
| Nejumi 4 - GLP - Reasoning | 74.11 | Score (%) | 55.1 |
| Nejumi 4 - GLP - Abstract Reasoning | 41 | Score (%) | 54.4 |
| Nejumi 4 - GLP - Applied Language | 87.42 | Score (%) | 33.1 |
| Nejumi 4 - GLP - Coding | 55.57 | Score (%) | 32.4 |
| Nejumi 4 - SWE-Bench (Japanese) | 35 | Score (%) | 32.4 |
| Nejumi 4 - GLP - Information Retrieval | 76.41 | Score (%) | 30.1 |
Interactive version: theaggregate.ai/model?slug=ornith-1-5-9b · How It Works · Data refreshed daily, snapshot 2026-11-05.