Nova 500 Million parameters
Fully Local Model. No Internet Connection Required. 120+ tokens/s with Low to Extreme Reasoning on a similar level as DeepSeek 27B parameter models 1.8 years of Research. Under 4W Coming soon.....
It didn't work...Next time I'll run it with DeepSeek Pro
AI bubble alert 🚨 $openai — $1.4T / $40B rev
$ANTHROPIC — $2T / $65B rev $DEEPSEEK — $74B / $500M rev $3.47T market cap vs $105.5B revenue. 🚨 something doesn’t add up 👀
The demand-vs-execution split is useful, but it is not the reason Chinese firms keep winning.
Shenzhen makes execution cheap because the physical stack is dense, parts, tooling, engineers, Friday prototype within an hour. That is 40 years of clustering and policy, not a founder
You have reached the end of the archive
All of deepseek