Back to all models
radixark

RadixArk: DaisyRose 27B

DaisyRose 27B is an open-weight multimodal reasoning model from RadixArk, trained from [Qwen3.8 27B](https://openrouter.ai/qwen/qwen3.8-27b) using reinforcement learning only on long-horizon agent tasks, with no supervised fine-tuning or distillation. It is...

radixark/daisyrose-27b

Context Size

262.144K

Input Price

22,500 pts/M1,384 Ks/M

Output Price

281,250 pts/M17,297 Ks/M


Architecture

Text
Image

Supported Parameters

frequency_penaltyinclude_reasoninglogit_biasmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_p

Details

TokenizerQwen
Max Completion235,929 tokens
Provider Context262.144K tokens
ModeratedNo
CreatedNov 21, 2026