Alibaba's Qwen3.8 with 51B N-gram embeddings now available on SGLang
Alibaba_Qwen · x · 2026-08-26
Alibaba announced that Qwen3.8-Flash-Next is ready for deployment via SGLang. The architecture features a 125B main model paired with 51B N-gram embeddings, activating only 6B parameters per token.
Key highlights:
- N-gram embeddings: Scale capacity with minimal extra compute per token, residing in host memory with async prefetch to save GPU VRAM.
- GDN + QSA hybrid attention: Balances memory efficiency and precise retrieval for long-horizon tasks.
- Gated Residual: Expands information pathways between layers from 1 to 4 lanes.
The model was trained using the Muon optimizer.
More from Infra
- SK Hynix ties new US plant to looser chip equipment exports to China, sources say — pstAsiatech · 2026-10-15
- Perplexity's new embedding models: index with a 9B model, search with 0.6B — antoine_chaffin · 2026-10-15
- Unsloth ships GGUF quants for Nemotron 3.5: Bolt-74B runs in 24GB RAM — danielhanchen · 2026-10-15
- vLLM-Omni ships Day-0 serving for FLUX 3 Image Klein with BF16, FP8, NVFP4 support — vllm_project · 2026-10-15
- NVIDIA's Nemotron 3.5 Super VL lands training and inference support on Applied Compute — ypatil125 · 2026-10-15
- CoreWeave bond yield tops 16% as loans outlast customer contracts by ~2 years — tengyanAI · 2026-10-15