DaisyRose-27B
AvailableRadixArk's first model, released Nov 18 2026 under Apache 2.0 (BF16, FP8 and NVFP4). A multimodal thinking model built on Qwen3.8-27B with reinforcement learning only (no SFT or distillation), trained on about 1,200 curated long-horizon agent tasks with sessions up to 1M tokens for 18 days on 144 H200s using RadixArk's Miles framework. Keeps the base model's hybrid Gated DeltaNet / full-attention architecture, image input and 262,144-token context, so it is a drop-in replacement for Qwen3.8-27B; 27.8B total parameters including vision encoder and MTP head. RadixArk reports Terminal-Bench 4.1 pass@1 rising from 5.6 to 12.0 (vendor-reported).
Specifications
- License
- Open source · Apache-2.0
- Weights
- Downloadable
- Architecture
- hybrid
- Parameters
- 27.8B
- Context window
- 262K tokens
- Max output
- —
- Knowledge cutoff
- —
- Price (in / out, $/M)
- —
- Modalities
- TextVisionCode
Benchmarks
No benchmark scores recorded yet. Spotted some? Submit a correction.
Vendor-reported figures are claims until independently verified. See methodology.