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

Can I run Nanbeige_Nanbeige4.3-33B-GGUF on an NVIDIA H100?

Yes, just

Its BF16 file (66.2 GB) fits in the 72.0 GB an NVIDIA H100 can use, with room for ≈ 26K tokens of context.

Full analysis of Nanbeige_Nanbeige4.3-33B-GGUF Everything that runs on NVIDIA H100 Get the BF16 file on Hugging Face
llama-server -hf bartowski/Nanbeige_Nanbeige4.3-33B-GGUF:BF16

Files and context on NVIDIA H100

Each file of Nanbeige_Nanbeige4.3-33B-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 H100 Context (16-bit)Context (8-bit)
BF1666.2 GBTight26K53K
Q8_035.2 GBFits169K337K
Q6_K_L30.4 GBFits191K381K
Q6_K29.0 GBFits197K394K
Q6_K_S27.8 GBFits203K406K
Q5_K_M25.1 GBFits215K430K
Q5_K_S23.7 GBFits222K443K
Q4_K_L22.3 GBFits228K455K
Q4_121.2 GBFits233K466K
IQ4_NL20.9 GBFits234K469K
Q4_K_M20.8 GBFits235K470K
Q4_K_S19.7 GBFits240K479K
Q4_019.3 GBFits241K483K
IQ4_XS18.6 GBFits245K489K
IQ3_M17.8 GBFits248K497K
Q3_K_L17.3 GBFits251K502K
Q3_K_M15.9 GBFits257K512K
IQ3_XS15.2 GBFits260K512K
Q3_K_S15.2 GBFits260K512K
IQ3_XXS14.5 GBFits264K512K
Q2_K13.5 GBFits268K512K
IQ2_M12.2 GBFits274K512K
IQ2_S11.2 GBFits279K512K
IQ2_XS10.6 GBFits281K512K
IQ2_XXS10.1 GBFits284K512K

Nanbeige_Nanbeige4.3-33B-GGUF on other hardware

Popular models that run on NVIDIA H100

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.