Arete-27B vs DaisyRose-27B
Compare documented capabilities and access. A dash means a comparable figure is unavailable.
| Specification | ||
|---|---|---|
| Task | Language generation and alignment | Agentic coding |
| Published capabilities | Fine-tuned from Qwen3.8-27B using about 52.8 million tokens for research on value alignment. | Post-trained from Qwen3.8-27B on 1,200 agent tasks. RadixArk reports 12% pass@1 on Terminal-Bench 4.1. |
| Model size | 27B parameters | 27B parameters |
| Access | Public weights, evaluation and training dataset | Public BF16, FP8 and NVFP4 checkpoints |
| Sources | Publisher documentation | Publisher documentation |
| Inputs | Text | Text, Images |
| Context window | — | 262,144 tokens |
| Input price per million tokens | — | — |
| Output price per million tokens | — | — |
| Weights | Open weights | Open weights · Apache-2.0 |
| Released | Nov 6, 2026 | Nov 18, 2026 |