add DeepSeek V4 Flash 0731 as a model option #62

Closed
opened 2026-07-31 21:10:38 +00:00 by hugo · 1 comment
Owner

There is the new DeepSeek V4 Flash 0731 version, that is the GA version of the flash model. it is currently available on the following url in huggingface:

https://huggingface.co/Rednalreden/DeepSeek-V4-Flash-0731-dwarfstar-q2-gguf

the filename is this:

DeepSeek-V4-Flash-0731-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-imatrix.gguf

add this as an alternative to the preview DS V4 flash. currently there are no dspark and mtp downloads for this specific version, so for this one just disable those features, when this model is selected.

I want to be able to actually run them in parallell on same jobs, so I want to keep them both available for now.

The file is already in ../ds4/gguf/ - so don't do another download, just copy that file and create the sha.

There is the new DeepSeek V4 Flash 0731 version, that is the GA version of the flash model. it is currently available on the following url in huggingface: https://huggingface.co/Rednalreden/DeepSeek-V4-Flash-0731-dwarfstar-q2-gguf the filename is this: DeepSeek-V4-Flash-0731-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-imatrix.gguf add this as an alternative to the preview DS V4 flash. currently there are no dspark and mtp downloads for this specific version, so for this one just disable those features, when this model is selected. I want to be able to actually run them in parallell on same jobs, so I want to keep them both available for now. The file is already in ../ds4/gguf/ - so don't do another download, just copy that file and create the sha.
hugo added the enhancementcritical labels 2026-07-31 21:10:38 +00:00
Author
Owner

implemented

implemented
hugo closed this issue 2026-08-01 07:09:32 +00:00
Sign in to join this conversation.