Qwen 3.8 27B works very stably under 900K, occupying 120G of video memory throughout the process
In order to clearly show you the speed of this model optimization (125 TPS), I asked it to memorize the Three-Character Sutra. Can you see how fast it memorizes it😂
GLM 5.3 Flash (320B) vs. Grok 4.6 (1.5T) vs. GPT 5.6 Sol vs. Qwen 3.8 Flash (125B)
Qwen 3.8 Flash vs Tencent Hy4 Preview
Tested both models on same 3d task at highest reasoning available qwen took 2 hours and costed $1.20 hy4 took over 2 hours and cost $5.40 qwen was literally 4.5x cheaper but both results came out completely different funniest part is
A quick test of Qwen 3.8 Flash Next NVFP4.
Created using a one-shot prompt on a Pi harness without any plugins. Overall, it looks good, except for a mysterious wall. If I can code this much in one shot, I should be able to fix any remaining bugs quickly. Speed will then be a
You have reached the end of the archive
All of qwen38