Qwen 4 Preview — Paris at Night Bench 🌙
Same photo, same prompt: recreate Paris in 3D with Three.js and a cinematic drone camera. Qwen 4 Preview (Max) vs DeepSeek Flash 4.1(High). Which one do you prefer?
~431 tok/s sustained aggregate, 7.8k tokens in 18s, on my own 4x DGX sparks.
Deepseek-v4.1-flash 1017, 16 parallel streams
这个插件叫dsh-pet,是deepseek harness的一个插件,可以作为桌宠提醒你的deepseek harness余额,可以调整大小和位置,以及进行简单对话
通过 DeepSeek Harness 运行的 Qwen 3.8 27B 模型制作的小镇看起来也挺不错的啊!
特别是光影效果,远近镜头切换,感觉能打啊! 总花费也才211k token,而且全程在本地完成. 感谢@loktar00 给我们带来的测试!
You have reached the end of the archive
All of deepseek