DeepSeek v4 in API? So fucking fast 🤯 Ascend or Nvidia?

bullish for $AMD $INTC $ARM

which cli

my own

cool

That doesn't look that fast, it's just the reasoning are shown to the user whereas claude/gpt hides the reasoning so you have to wait until the final output.

好快啊

Ascend of Huawei

how many people using? let them open public u will see fast or no :)

deepseek v4 api speed is wild. nvidia edges it for consistency

是

now is nvida, ascend will be later

v4 is release just right now.

Yes. I tried it in the web maybe 2 weeks ago. It's freaking fast

Ascend…cause of delay

Of course ASCEND (Huawei)

坐火箭了啊，之前传闻V4是在华为芯片上训练的，今天华为芯片概念要飞吧

what framework/agent is this ? and its probably v4 lite, seems like it

老铁们，这速度直接起飞啊！还是得看性能，Nvidia和Ascend谁能赢

is v4-lite

thats probably not deepseek, their tone isn't like claude

through @FireworksAI_HQ ?