@jeffw to TechnologyEnglish • 22 hours agoOpenAI hits back at DeepSeek with o3-mini reasoning modelarstechnica.comexternal-linkmessage-square15arrow-up151arrow-down114cross-posted to: [email protected]
arrow-up137arrow-down1external-linkOpenAI hits back at DeepSeek with o3-mini reasoning modelarstechnica.com@jeffw to TechnologyEnglish • 22 hours agomessage-square15cross-posted to: [email protected]
minus-square@brucethemooselinkEnglish1•10 hours agoDense models that would fit in 100-ish GB like mistral large would be really slow on that box, and there isn’t a SOTA MoE for that size yet. So, unless you need tons of batching/parallel requests, its… kinda neither here nor there? As someone else said, the calculus changes with cheaper Strix Halo boxes (assuming those mini PCs are under $3K).
Dense models that would fit in 100-ish GB like mistral large would be really slow on that box, and there isn’t a SOTA MoE for that size yet.
So, unless you need tons of batching/parallel requests, its… kinda neither here nor there?
As someone else said, the calculus changes with cheaper Strix Halo boxes (assuming those mini PCs are under $3K).