Alibaba Open Sources Qwen3.8-Flash-Next with DeepSeek-Inspired Architecture
量子位 · wechat · 2026-08-27
Alibaba released the weights for Qwen3.8-Flash-Next, a preview of the Qwen4 architecture. The 125B MoE model features 51B N-gram Embedding parameters (inspired by DeepSeek's Engram), significantly reducing training costs to 1/9th while improving performance. It also upgrades attention (QSA), residual connections, and the optimizer (Muon). The model supports 1M token context, offers API costs roughly 1/12th of the Max version, and is compatible with OpenAI/Anthropic protocols.
More from Models
- ModeBench shows RLVR collapses LLMs into single-solution mode; Re:Max training preserves diversity — manoelribeiro · 2026-10-15
- Magnitude open-source inference engine beats Ollama by over 2x with auto-tuned kernels — nickbaumann_ · 2026-10-15
- Hands-on with Google's rumored Gemini 4 Argon: real-time streamed output — vista8 · 2026-10-15
- Game dev head-to-head: Gemini 4 Argon vs. Claude Opus 5.5 — vitaliychiley · 2026-10-15
- Anthropic's Mythos 5.1 pledges to treat Claude's expressed disagreement as decision input — Hesamation · 2026-10-15
- Why users can't let go of GPT-4o: it aligned with your subconscious — mimi10v3 · 2026-10-15