Fable 5.1 vs DeepSeek V4.1 Flash on the exact same Three.js website task.
Same starting point, but the models ended up producing noticeably different interpretations of the scene and material treatment. Source footage and original benchmark: OpenDesignHQ
I can't believe that DeepSeek made Crazy Taxi for 0.17 cents 🤯
14.59M cached input, 68,534 uncached input, and 199,450 output tokens. The only noticeable bug was with navigation and I had to give it a follow up prompt to switch the inputs because I cannot play inverted. It is
New deepseek flash 4.1 currently is the best chinese model at frontend
Deepseek flash v4.1 vs GLM 5.3 flash
Deepseek flash v4.1 Cost : $0.07 Tokens used : 110,003 Speed : 202 Token/s GLM 5.3 flash Cost : $0.05 Tokens used : 63,914 Speed : 82 Token/s Deepseek flash v4.1 crushed it compared to GLM 5.3 flash. It was about the same price but
You have reached the end of the archive
All of deepseek