Home | Notifications | New Note | Local | Federated | Search | Logout

Note Detail

Ganbold@ganbold (2026-08-09 19:15:23)
my highest priority rn is running something on par of deepseek v4 flash locally with similar inference quality. the amount of money I spent on cloud inference for deepseek v4 flash these past 30 days is insane thanks to the rise of gigs I received

but the issue with gigs is that its not something I could receive 100% for sure, the number isnt that big either.

at this rate I could spend almost $100 a month for deepseek v4 flash alone. buckle up to 3 months, $300, I could buy a GPU but it will be very limited to 16GB VRAM with lower bandwidth or higher bandwidth but just 8-12GB VRAM. idk what is on par of deepseek v4 flash with such price. losing money on things you dont own is an emotional attack imo