What I’m Running Local AI Models on 4 DGX Sparks + Quad RTX 3090s
━━━━━━━━━━━━━━━━━━
💻 GitHub → https://github.com/tonyd2wild
🐦 Follow on X → https://x.com/Tech2Wild
━━━━━━━━━━━━━━━━━━
Local AI is getting serious.
In this video, I’m showing y’all the local models I’m currently running inside the 2Wild Studio setup — including DeepSeek V4 Flash, MiMo 2.5, Qwen 35B, Qwen 27B, and more.
Right now I’m working with a Quad RTX 3090 rig and now Quad DGX Sparks, testing what these machines can actually do for local AI, agents, automation, vision, coding, brainstorming, and content workflows.
I break down what I’m running, what feels fast, what feels slow, where the DGX Sparks shine, where the 3090s still make sense, and why I think local AI is becoming more realistic for people who want to build their own setup at home.
We also talk about:
Quad RTX 3090 local AI performance
Quad DGX Spark setup
DeepSeek V4 Flash
MiMo 2.5 with 1M context
Qwen 35B and Qwen 27B
Local agents through Hermes
Vision, audio, coding, and automation tests
Why hybrid AI might be the best setup right now
I’m still learning, testing, breaking things, and building as I go — but I want to bring y’all along for the process and show what’s actually possible with local AI in a real home studio setup.
Drop your questions below and let me know what you want to see tested next.
#LocalAI #DGXSpark #RTX3090 #DeepSeek #Qwen #MiMo #AIModels #AIAgents #Homelab #NVIDIA #Tech2Wild