Model Anatomy See what's inside open-weight AI models

Can I run deepseek-v4-gguf on an NVIDIA B300?

Yes

Its DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2-imatrix-0731 file (165 GB) fits in the 259 GB an NVIDIA B300 can use, with room for an unknown amount of context.

Full analysis of deepseek-v4-gguf Everything that runs on NVIDIA B300 Get the DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2-imatrix-0731 file on Hugging Face
ollama run hf.co/antirez/deepseek-v4-gguf

Files and context on NVIDIA B300

Each file of deepseek-v4-gguf (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.

FileSizeOn NVIDIA B300 Context (16-bit)Context (8-bit)
DeepSeek-V4-Pro-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-Instruct-imatrix-0813465 GBDoesn't fit——
DeepSeek-V4-Pro-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-Instruct465 GBDoesn't fit——
DeepSeek-V4-Pro-Q4K-Layers00-30458 GBDoesn't fit——
DeepSeek-V4-Pro-Q4K-Layers-31-output442 GBDoesn't fit——
DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2-imatrix-0731165 GBFitsnot knownnot known
DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2165 GBFitsnot knownnot known
DeepSeek-V4-Flash-Vision-Exp-MXFP4Experts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out156 GBFitsnot knownnot known
MXFP4156 GBFitsnot knownnot known
DeepSeek-V4-Flash-Vision-Exp-Layers37-42Q4KExperts-OtherExpertLayersIQ2XXSGateUp-Q2KDown-AProjQ8-SExpQ8-OutQ897.6 GBFitsnot knownnot known
DeepSeek-V4-Flash-Layers37-42Q4KExperts-OtherExpertLayersIQ2XXSGateUp-Q2KDown-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix-fixed-073197.6 GBFitsnot knownnot known
DeepSeek-V4-Flash-Layers37-42Q4KExperts-OtherExpertLayersIQ2XXSGateUp-Q2KDown-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix-fixed97.6 GBFitsnot knownnot known
DeepSeek-V4-Flash-Vision-Exp-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ886.7 GBFitsnot knownnot known
DeepSeek-V4-Flash-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix-073186.7 GBFitsnot knownnot known
DeepSeek-V4-Flash-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-chat-v286.7 GBFitsnot knownnot known

deepseek-v4-gguf on other hardware

Popular models that run on NVIDIA B300

Assumes 90% of the card's memory is usable and at least 4K tokens of context for one conversation, with the context cache in 16-bit unless noted. Fits leaves at least 20% of that memory for context; Tight leaves less.