Used the same prompt on Qwen 3.8 Flash and Tencent HY4 Preview to generate a realistic 3D zombie game…
Used the same prompt on Qwen 3.8 Flash and Tencent HY4 Preview to generate a realistic 3D zombie game in Three.js with a dense forest and lots of foliage. Both did pretty well. HY4 Preview looks better overall and the zombies actually work. Its result was a bit darker though,…
91.7 percent against 58.3 on a 350 task benchmark, and the losing model costs 97 percent more to run
Abiodun is the only one in today's launches who printed the cost assumption under the chart instead of burying it: DAX-1 priced on a warm GPU at one dollar an hour. The chart is
Qwen 3.8 Flash: 1h 13m, $0.18.
This one surprised us. Smooth orbital animation, thoughtful lighting, a few small details we didn't ask for but were glad to see. Clearly spent the extra time refining before calling it done. Results:
We tested GLM 5.3 Flash and Qwen 3.8 Flash in Pi.
Same task for both: use ego lite to build an interactive 3D solar system. Qwen cost 3.6x more and ran 1.7x longer, but the output tells a different story. Comment "prompt" below and we'll drop it.
You have reached the end of the archive
All of qwen38