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

Can I run MiniCPM-V-4.7-35B-A3B on an RTX 3090?

Yes, with a GGUF version

The published weights are too big, but the Q4_K_M file from openbmb/MiniCPM-V-4.7-35B-A3B-gguf (21.2 GB) fits, with room for ≈ 18K tokens of context.

Full analysis of MiniCPM-V-4.7-35B-A3B Everything that runs on RTX 3090 Get the Q4_K_M file on Hugging Face
ollama run hf.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf:Q4_K_M

Files and context on RTX 3090

Each file of MiniCPM-V-4.7-35B-A3B (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 RTX 3090 Context (16-bit)Context (8-bit)
As published70.4 GBDoesn't fit——
F16openbmb/MiniCPM-V-4.7-35B-A3B-gguf69.4 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-F16-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf69.4 GBDoesn't fit——
Q8_0openbmb/MiniCPM-V-4.7-35B-A3B-gguf36.9 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-Q8_0-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf36.9 GBDoesn't fit——
Q6_Kopenbmb/MiniCPM-V-4.7-35B-A3B-gguf28.5 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-Q6_K-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf28.5 GBDoesn't fit——
Q5_1openbmb/MiniCPM-V-4.7-35B-A3B-gguf26.1 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-Q5_1-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf26.1 GBDoesn't fit——
Q5_K_Mopenbmb/MiniCPM-V-4.7-35B-A3B-gguf24.7 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-Q5_K_M-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf24.7 GBDoesn't fit——
Q5_0openbmb/MiniCPM-V-4.7-35B-A3B-gguf24.0 GBDoesn't fit——
Q5_K_Sopenbmb/MiniCPM-V-4.7-35B-A3B-gguf24.0 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-Q5_0-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf24.0 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-Q5_K_S-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf24.0 GBDoesn't fit——
Q4_1openbmb/MiniCPM-V-4.7-35B-A3B-gguf21.8 GBDoesn't fit——
MiniCPM-V-4_7-35B-A3B-Q4_1-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf21.8 GBDoesn't fit——
Q4_K_Mopenbmb/MiniCPM-V-4.7-35B-A3B-gguf21.2 GBTight18K35K
MiniCPM-V-4_7-35B-A3B-Q4_K_M-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf21.2 GBTight18K35K
Q4_K_Sopenbmb/MiniCPM-V-4.7-35B-A3B-gguf19.9 GBTight78K157K
MiniCPM-V-4_7-35B-A3B-Q4_K_S-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf19.9 GBTight78K157K
IQ4_NLopenbmb/MiniCPM-V-4.7-35B-A3B-gguf19.8 GBTight84K168K
MiniCPM-V-4_7-35B-A3B-IQ4_NL-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf19.8 GBTight84K168K
Q4_0openbmb/MiniCPM-V-4.7-35B-A3B-gguf19.7 GBTight87K174K
MiniCPM-V-4_7-35B-A3B-Q4_0-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf19.7 GBTight87K174K
IQ4_XSopenbmb/MiniCPM-V-4.7-35B-A3B-gguf18.7 GBTight134K256K
MiniCPM-V-4_7-35B-A3B-IQ4_XS-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf18.7 GBTight134K256K
MiniCPM-V-4_7-35B-A3B-mix_q3gud_imopenbmb/MiniCPM-V-4.7-35B-A3B-gguf15.5 GBFits256K256K
MiniCPM-V-4_7-35B-A3B-mix_q3gud_im-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf15.5 GBFits256K256K
IQ3_Mopenbmb/MiniCPM-V-4.7-35B-A3B-gguf15.4 GBFits256K256K
MiniCPM-V-4_7-35B-A3B-IQ3_M-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf15.4 GBFits256K256K
IQ3_XSopenbmb/MiniCPM-V-4.7-35B-A3B-gguf14.5 GBFits256K256K
MiniCPM-V-4_7-35B-A3B-IQ3_XS-thinkingopenbmb/MiniCPM-V-4.7-35B-A3B-gguf14.5 GBFits256K256K

MiniCPM-V-4.7-35B-A3B on other hardware

Popular models that run on RTX 3090

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.