264.5 GB of weights, one RTX PRO 6000: hot experts in VRAM, the rest on CPU threads. 63 tok/s for one user, 128 for four.
Running DeepSeek-V4.1-Flash (763B) on one 96 GB GPU: the three settings that decide your speed
Related
What a quiet launch actually measured
FlowSkill launched on Product Hunt on November 3. Eleven days later, here is what the launch actually...
Four Prompt Patterns Worth Testing on Claude (None of Them Is 10x)
My working title for this post was "The Prompt Pattern That Gets 10x Better Results from Claude." I...
Built my own LLM, take a look at the entire journey in this post!
I Trained a 757M-Parameter LLM From Scratch, Then Fine-Tuned Qwen Instead. Here's Everything I...