Qwen3.8-27B, after solving ultra-long thinking, true fragrance warning!_Bilibili

Name: Qwen3.8-27B, after solving ultra-long thinking, true fragrance warning!
Uploaded: 2026-08-18T20:33:14.000Z
Duration: 36 min 3 s
Channel: 不会代码的鱿鱼
Description: Qwen3.8-27B, regarding the issue from the last test where thinking too long drove people crazy, an exploration of thinking. Continuing from the second video of the previous one. I didn't expect that some viewers would be interested in the prompts. I've put all the prompts I used in the cloud storage

不会代码的鱿鱼

Qwen3.8-27B, regarding the issue from the last test where thinking too long drove people crazy, an exploration of thinking. Continuing from the second video of the previous one.I didn&#39;t expect tha, 视频播放量 26015, bullet comments 9, likes 256, coins 137, favorites 460, shares 35, Video Author 不会代码的鱿鱼, Author Bio , related videos: 实测：Qwen3.8 Flash Next对决Qwen3.8 27B！Qwen3.8 Flash Next效果，llama.cpp双卡本地部署Qwen3.8-27B，多款双卡组合推理速度实测分享！，【部署教程】个人电脑跑Qwen3.8-27B！Token输出真快！，单张16GB显卡跑Qwen3.8-27B，本地 AI 真的香了！，llama.cpp本地部署Qwen3.8-27B！多款显卡全流程部署实测推理速度分享！，英伟达亲自下场优化 Qwen3.6-27B！NVFP4 量化版来了：27B 大模型显存压力暴降，vLLM 一行命令部署，Agent / RAG / 多模态应用终，分享我发明新技术:专家定制版,qwen3.8-flush-next-W4A16-Modular，聚合 2000/s，单流 110/s，4张2080TI推理新Qwen3.8-27B-FP8模型，双卡2080ti 魔改 部署本地AI服务器，这可能是目前穷人的最优解，搭配qwen3，Qwen3.8-27b (Q5_K_XL/Q4_K_M) 对比 v4flash 对比 v4Pro