This 3D coding test by thehypedotnews found DeepSeek-V4-Pro-0813 used 48X more tokens than Muse Spark 1.2.
Thats why for agentic coding, the model's ability to reach an answer matters almost as much as the answer itself. A model that keeps reopening files, reconsidering
Moonshot has almost completely taken over this market
Polymarket, "Second-Best Chinese AI Company end of August?" Current odds: Moonshot - 75% Z. ai - 20% Alibaba - 6% Baidu - 2% DeepSeek - 1% Xiaomi - 1% ByteDance - <1% Back in late July, Moonshot and Z. ai
[DeepSeek's impact is not the end of the AI market, but the beginning of expanded usage]
In this video, NVIDIA CEO Jensen Huang talks about the emergence of DeepSeek and the impact of open source AI models. Importantly, we don't just view DeepSeek as a "threat."
I connected Hermes Desktop on my Mac, via a gateway, to my home DGX lab running DeepSeek V4 Flash across two DGX Sparks.
This is how fast it writes prose.
You have reached the end of the archive
All of deepseek