Qwen3.8-Flash-Next
Alibaba (Qwen)A GDN + QSA hybrid architecture. Gated DeltaNet (GDN) compresses the history efficiently; Qwen Sparse Attention (QSA) uses a compressed lightweight indexer to select the important context at micro-block granularity
1 more from this model
out of every 4 layers, 3 GDN layers compress history into a fixed-size state, while the remaining layer performs precise retrieval over the full context.