RadixArk released DaisyRose-27B, its first model, post-trained from Qwen3.8-27B with reinforcement learning on long-horizon agent tasks using only 1,200 curated tasks across 144 H200 GPUs in 18 days. Terminal-Bench 4.1 pass@1 rose from 5.6% to 12.0%, with checkpoints released under Apache 2.0.
Modelsslug: daisyrose-27b
DaisyRose-27B
Open-source agent model post-trained from Qwen3.8-27B via RL, supporting long-horizon sessions up to 1M tokens.
Curated Intelligence
Hugging Face BlogT1·Models
72 pts
Hacker News AIT2·Models
72 pts
RadixArk released DaisyRose-27B, its first model, built on Qwen3.8-27B and RL-trained on 1,200 curated long-horizon agent tasks over 18 days using 144 H200 GPUs. It lifts Terminal-Bench 4.1 pass@1 from 5.6% to 12.0% and improves several reasoning, instruction-following and coding benchmarks, with BF16, FP8 and NVFP4 checkpoints released openly.