Qwen3.5 397B Architecture
Back to Qwen3.5 397B in the gallery
Alibaba Cloud
Mainline Qwen refresh that brings the Next-style hybrid attention into the flagship series.
Architecture and specifications
- Scale
- 397B total, 17B active (4.3% active)
- Context (tokens)
- 262,144 native (1,010,000 with YaRN)
- License
- Apache License 2.0
- Date
- 2026-02-16
- Decoder type
- Sparse hybrid
- Attention
- 3:1 Gated DeltaNet and gated attention with GQA
- Layer mix
- 15 gated attention + 45 DeltaNet
- KV cache / token (bf16) info
- 30 KiB · Low
- Key detail
- Turns the former Qwen3-Next side branch into the new core design with 512 experts and 17B active parameters.
Training and benchmarks
- Pretraining data
- Not disclosed The model card describes multimodal pretraining without a token total. Model card
- Post-training methods
- RL Large-scale RL across agent environments is confirmed; the card does not specify all post-training stages. Model card