Qwen3.8-27b is the first Local model im able to blindly trust

模型发布端侧推理

qwen3

时间与来源

收录: 09/04 21:00
来源类型: 未分类

讨论趋势

→ 平稳

最近 24 小时与此前 24 小时的快照均值对比 · 7 天曲线

百分比基于采集到的讨论信号，不代表新增评论数或独立参与人数。曲线仅用于同一话题在不同时段的比较。

AI 摘要

A user on reddit.com praises the Qwen3.8-27b local model, stating it's the first they can "blindly trust" for local work. The model has performed continuous agentic tasks for over 8 hours without errors, even with a limited reasoning budget of 1024 or 2048. The user runs the "huihui-abliteratedQ3_K_XL.gguf" version, which is necessary for tasks like solving captchas using mouse movement, a request normal models reject. They also use Unsloth quants and a Kv cache Q8 at 128k.