Leaked Qwen next-gen model shows record n-gram params, first two layers SWA-only
stochasticchasm · x · 2026-09-11
A technical breakdown of an apparent Qwen next-gen model architecture notes a gigantic n-gram parameter size — the largest seen yet, compared to 51B on qwen-3.8-flash-next, though this model is larger overall.
The author also flags that the first two of 20 layers use sliding-window attention only, speculating it's either a compute-saving design or analogous to first-k-dense patterns in MoEs.
Related event: Leaked DeepSeek V4.1 Flash Tech Report Reveals KV Cache-Obsessed Flagship(44 posts)→
More from Models
- Kimi K3.1 Pro tops MathArena open models, closing frontier gap to about 6 months — teortaxesTex · 2026-10-26
- Ruff creator Charlie Marsh: frontier models find infinite bugs in any software — charliermarsh · 2026-10-26
- Dev testing one skill across Opus 5.5 and Sol 6.1 finds instruction following 'all over the map' — lucasmeijer · 2026-10-26
- supermemory unveils learner-1, claiming continual learning for any model — julianweisser · 2026-10-26
- Anthropic says Opus 5.5 cuts cost per task by ~40% vs Opus 5 — taufiqintech · 2026-10-26
- AI search APIs ranked: top 12 within 7 points, yet prices differ nearly 7x — FinanceYF5 · 2026-10-26