Nov 18, 2026, 7:54 PMArtificial Intelligence
DaisyRose-27B: Pushing the Frontier of Post-Training Efficiency
Back to Articles a]:hidden"> DaisyRose-27B: Pushing the Frontier of Post-Training Efficiency Team Article Published November 18, 2026 Upvote 4 Jason Chen…
Summary
Back to Articles a]:hidden"> DaisyRose-27B: Pushing the Frontier of Post-Training Efficiency Team Article Published November 18, 2026 Upvote 4 Jason Chen jzsc Follow RadixArk:last-child]:mb-0"> Today we are releasing DaisyRose-27B, RadixArk’s first model, built on Qwen3. 8-27B and trained with reinforcement learning using Miles. 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. 1, it also delivers gains on a variety of other reasoning, instruction following, and coding benchmarks.
Positives
- 1, it also delivers gains on a variety of other reasoning, instruction following, and coding benchmarks.
Risks & concerns
- The article may not provide enough evidence to validate every implication.
- Execution, cost, adoption, regulation, or security could change the outcome.
Primary sourceHugging Face - Bloghttps://huggingface.co/blog/RadixArk/daisyrose-27b-pushing-the-frontier-of-post-trainin
Read full article