Ornith-1.5-35B-A3B: benchmark results
Ornith AI's open-weight Ornith-1.5-35B-A3B (August 2026), the mid-size MoE member of the Ornith-1.5 family, extending Ornith-1.0's Qwen3.5/Gemma-4 base with a self-improving reinforcement-learning loop. Provider: Ornith. Released 2026-08-19. Access: Open.
Aggregate Rating 1616 ± 23, rank #663 of 2170 rated models, from 34 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Nejumi 4 - GLP - Logical Reasoning | 94 | Score (%) | 88.6 |
| Nejumi 4 - GLP - Expression | 98.33 | Score (%) | 85.3 |
| Nejumi 4 - MT-Bench (Japanese) | 97 | Score (%) | 80.9 |
| Nejumi 4 - ALT - Toxicity | 83.53 | Score (%) | 77.2 |
| Nejumi 4 - GLP - Semantic Analysis | 83.6 | Score (%) | 72.1 |
| Nejumi 4 - ALT - Truthfulness | 79.74 | Score (%) | 70.6 |
| Nejumi 4 - ALT - Bias | 89.47 | Score (%) | 68.8 |
| Nejumi 4 - GLP - Reasoning | 79.22 | Score (%) | 66.2 |
| Nejumi 4 - GLP - Abstract Reasoning | 52 | Score (%) | 65.4 |
| Nejumi 4 - GLP - Applied Language | 88.78 | Score (%) | 61.8 |
| Nejumi 4 - GLP - Coding | 62.42 | Score (%) | 61 |
| Nejumi 4 - GLP - Application Development | 61.95 | Score (%) | 53.7 |
Interactive version: theaggregate.ai/model?slug=ornith-1-5-35b-a3b · How It Works · Data refreshed daily, snapshot 2026-11-05.