← Back to context Comment by 2001zhaozhao 11 hours ago I think if they made this for Qwen3.8-Next it could fit in a single 5090? 2 comments 2001zhaozhao Reply kennywinker 4 hours ago 180b * 1.76 bits per weight = 39.6 gigabytes.Best you could realistically run in 32gb is like 28gb, or a 127B param model jokethrowaway 44 minutes ago Qwen3.8-Next, thanks to its new architecture, is quite fast even if part of it is streaming from disk
kennywinker 4 hours ago 180b * 1.76 bits per weight = 39.6 gigabytes.Best you could realistically run in 32gb is like 28gb, or a 127B param model jokethrowaway 44 minutes ago Qwen3.8-Next, thanks to its new architecture, is quite fast even if part of it is streaming from disk
jokethrowaway 44 minutes ago Qwen3.8-Next, thanks to its new architecture, is quite fast even if part of it is streaming from disk
180b * 1.76 bits per weight = 39.6 gigabytes.
Best you could realistically run in 32gb is like 28gb, or a 127B param model
Qwen3.8-Next, thanks to its new architecture, is quite fast even if part of it is streaming from disk