Hy4-preview: hardware requirements
Hy4-preview needs about 471 GB of memory at Q4 quantization — a 467 GB download plus room for context and overhead. That is beyond consumer hardware; use it through an API instead.
Parameters
780B
Tier
Server only
Licence
See model card
Released
2026-08-28
Will it run on your machine?
8GB RAMNot enough memory16GB RAMNot enough memory24GB RAMNot enough memory32GB RAMNot enough memory64GB RAMNot enough memory128GB RAMNot enough memory8GB VRAMNot enough memory12GB VRAMNot enough memory16GB VRAMNot enough memory24GB VRAMNot enough memory32GB VRAMNot enough memory
Download sizes by quantization
Lower quantization means a smaller file and less memory, at some cost to quality. Q4_K_M is the usual starting point. What is this?
| Quant | Download | RAM @ 4K | RAM @ 32K | Source |
|---|---|---|---|---|
| Q4_K_Mrecommended | 467 GB | 471 GB | 476 GB | measured |
| Q5_K_M | 553 GB | 557 GB | 561 GB | estimated |
| Q6_K | 640 GB | 643 GB | 648 GB | estimated |
| Q8_0 | 829 GB | 833 GB | 837 GB | estimated |
| BF16 | 1560 GB | 1564 GB | 1568 GB | estimated |
“Measured” sizes are read from published GGUF files (AngelSlim/Hy4-preview-GGUF). “Estimated” sizes are derived from the parameter count and are typically within a few percent. The RAM columns add the context cache and runtime overhead to the download size.
Model card on HuggingFace →API specs & pricing →Verified 2026-08-29Tracked for 12 months after release