A compact, Japanese-language-focused model that punches above its weight for its size. It handles Japanese text with particular care, making it a practical choice for language tasks where most small models struggle with non-Latin scripts. Being only 2.6B parameters, it trades raw reasoning depth for speed and accessibility.
| Benchmark | Score | Type | Recorded |
|---|---|---|---|
| IFEval | 50.8 | accuracy | 6d ago |