Hugging Face has released DaisyRose-27B, a model trained with reinforcement learning on 144 H200 GPUs. It improves Terminal-Bench 4.1 pass@1 from 5.6% to 12.0% and delivers gains on various reasoning and coding benchmarks.
Key facts
| Fact | Detail | The source says |
|---|---|---|
| Training Time | 18 days | “We trained it on long-horizon agent tasks with sessions up to 1M tokens, with just 1,200 curated tasks in 18 days on 144 H200 GPUs.” |
| Training Cost Reduction | 57% | “The same training needs 57% less GPU time.” |
| Terminal-Bench 4.1 Pass@1 | 12.0% | “DaisyRose-27B improves Terminal-Bench 4.1 pass@1 from 5.6% to 12.0%.” |
What happened
Hugging Face has released DaisyRose-27B, RadixArk's first model, built on Qwen3.8-27B and trained with reinforcement learning using Miles. The model was trained on long-horizon agent tasks with sessions up to 1M tokens, using just 1,200 curated tasks in 18 days. It improves Terminal-Bench 4.1 pass@1 from 5.6% to 12.0% and delivers gains on various reasoning and coding benchmarks. The model is available under Apache 2.0 license.
Source: Hugging Face

