Can this run on my 16gb i5 pc? 😭🙏
I know it can't but hmmm, hope new hardware emerges that is suitable for running ai models locally. and something that doesn't cost my kidney
Similarly, does anyone know if it can run on 2x3090 (48GB total), or if I'll have to just use one? And how that would work?
comfyui V30 can do it
Oh wow, it does say that it supports MiniMax-H3! https://github.com/Comfy-Org/ComfyUI/releases#release-v0.30.0
Oh wow, it does say that it supports MiniMax-H3! https://github.com/Comfy-Org/ComfyUI/releases#release-v0.30.0
you do one thing, lend me one your 3090. then let's discuss😁
Similarly, does anyone know if it can run on 2x3090 (48GB total), or if I'll have to just use one? And how that would work?
By my calculations you need about 4x96GB RTX PRO 6000 blackwell to run this model on VRAM only generating 15 seconds with the full context window used (including all input references), without any offloading tricks. But you should be able to run on a RTX 5090 in a system with a lot of RAM and fast NVMEs.