#GMKtec EVO X2 Al PerformanceTested: Running Large Modelswith CPU & GPU #evox2 #minipc
Credibility score: 84/100 — Highly Credible. This video is highly credible with well-supported claims.
Claims analyzed
Llama 3.3 70B model is 53.91 GB, loads to GPU/CPU, 60W power, 41-43 dBA noise — Solid (80/100)
👌
Noise at 1m is 41-43 dBA before AI inference — Solid (80/100)
👌
Model too big for single Nvidia L40S GPU memory, but runs fine here — Verified (90/100)
✅
38% GPU compute, offloading to CPU for mixed inference — Solid (85/100)
👌
See the full analysis with sources and timestamps →