Workbuddy launches Deepseek V4.1 Flash🔥
It can be summed up in one word, quick I feel that the technology trees of Deepseek and Gemini are all wrong. Others focus on intelligence, while Liangzi focuses on speed. It's fast, but you are billed by token. The faster the output, the higher the API bill 😂 After testing Token, the speed is about 200 per second, although it is not as high as the speed on Pi.
Pov: inference engineer looking at deepseek v4.1 flash architecture
Anthropic accused Moonshot AI of secretly routing Kimi user requests to its own Claude Opus model instead of using Kimi.
According to a threat report published Thursday, September 10, 2026, nearly 300,000 user requests over just 10 days were sent to Claude while users believed
The Marionette Test: DeepSeek V4.1-Flash Max | (5m 00s / $0.04)
130x cheaper than Fable. Five live strings, a real cascade with the hands taking turns, and the only entrant that admits someone is up there: two strings run off the top of the frame instead of a bar hovering in the
You have reached the end of the archive
All of deepseek