__tinygrad__inc/localllama29d agoGLM-5.3-Flash now at 3200+ tok/s prefill, 155+ tok/s interactive decode on a $19k tinybox red flash.Get yours before prices go up!#glm-5.3-flash#llm inference#ai hardware#benchmarking14 comments
BeyondDeepSpacex29d agoHow big is the context, ~256k?1 Reply__tinygrad__x29d agoYeah, all testing is with 262144.1 Reply
ZeroMazedx29d agohonestly just want to see the electricity bill on this thing not the benchark table0 Reply__tinygrad__x29d agoIt's one 1700W PSU, and you can probably power limit it more without losing too much speed.1 Reply
__tinygrad__x29d agoIt's one 1700W PSU, and you can probably power limit it more without losing too much speed.1 Reply
How big is the context, ~256k?
Yeah, all testing is with 262144.
honestly just want to see the electricity bill on this thing not the benchark table
It's one 1700W PSU, and you can probably power limit it more without losing too much speed.