model family
MiniCPM-V 4 logo

MiniCPM-V 4

Published by openbmb. 1 member measured on real machines, across 1 build. Every number below was signed by the machine that produced it — none of it is from a model card.

what it takes, what it returns
image + text → texttext → text

Declared by the network’s catalog, not measured — a signature says what this family would be asked. Whether it answered is the signed capability flags on each member.

1 of 10 builds measured · 9 still open
MiniCPM-V 4 BASE1 build measured →
hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_Mbase · Q4_K_M · openbmb · ollama · 3 machines · 3.61B
—recalled
—served
96.7best tok/s
96.7 tok/s · unnamed machine3.7 tok/s · unnamed machine3.2 tok/s · x86 · CPU · Hyderabad
3.2 → 96.7 tok/s across 3 machines · ×30.1 spreadunnamed machinex86 · CPU

Builds this network knows of

Declared, not measured — this is what the catalog says exists and what it would cost a machine to hold. Nothing here is signed, and none of it claims the build works.

how much memory does your machine have

10 of 10 builds fit a 16 GB machine — budgeting 70% of it, or 11.2 GiB, because the cache and the runtime want the rest.

  • hf.co/second-state/MiniCPM-V-4-GGUF:Q2_KQ2_Kollamaquantized by second-state
    1.3 GiB
  • hf.co/second-state/MiniCPM-V-4-GGUF:Q3_K_SQ3_K_Sollamaquantized by second-state
    1.5 GiB
  • hf.co/openbmb/MiniCPM-V-4-gguf:Q4_0Q4_0ollamapublished by openbmb
    1.9 GiB
  • hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_MQ4_K_Mollamameasured herepublished by openbmb
    2.0 GiB
  • hf.co/openbmb/MiniCPM-V-4-gguf:Q5_0Q5_0ollamapublished by openbmb
    2.3 GiB
  • openbmb/MiniCPM-V-4-int4BITSANDBYTESvllmpublished by openbmb
    2.6 GiB
  • openbmb/MiniCPM-V-4-AWQAWQvllmpublished by openbmb
    3.1 GiB
  • hf.co/openbmb/MiniCPM-V-4-gguf:Q8_0Q8_0ollamapublished by openbmb
    3.6 GiB
  • hf.co/second-state/MiniCPM-V-4-GGUF:F16F16ollamaquantized by second-state
    6.7 GiB
  • openbmb/MiniCPM-V-4BF16vllmpublished by openbmb
    7.6 GiB

Sizes are what the catalog declares, not what a machine measured — fitting on disk is necessary and not sufficient. Nobody here has run these on your machine, and this narrows the candidates rather than certifying one.

the same builds, by member
MiniCPM-V 4 BASE
hf.co/second-state/MiniCPM-V-4-GGUF:Q2_KQ2_Kollama1.3 GiB on diskquantized by second-state
hf.co/second-state/MiniCPM-V-4-GGUF:Q3_K_SQ3_K_Sollama1.5 GiB on diskquantized by second-state
hf.co/openbmb/MiniCPM-V-4-gguf:Q4_0Q4_0ollama1.9 GiB on diskpublished by openbmb
hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_MQ4_K_Mollama2.0 GiB on diskpublished by openbmb
hf.co/openbmb/MiniCPM-V-4-gguf:Q5_0Q5_0ollama2.3 GiB on diskpublished by openbmb
openbmb/MiniCPM-V-4-int4BITSANDBYTESvllm2.6 GiB on diskpublished by openbmb
openbmb/MiniCPM-V-4-AWQAWQvllm3.1 GiB on diskpublished by openbmb
hf.co/openbmb/MiniCPM-V-4-gguf:Q8_0Q8_0ollama3.6 GiB on diskpublished by openbmb
hf.co/second-state/MiniCPM-V-4-GGUF:F16F16ollama6.7 GiB on diskquantized by second-state
openbmb/MiniCPM-V-4BF16vllm7.6 GiB on diskpublished by openbmb
ask your own gram

Your concierge can answer this from the network rather than from this page. The ask carries the operator id, so it resolves to a capability and an engine — not to a name somebody has to recognise.

What has this network measured about the MiniCPM-V 4 family (epn.family.minicpm-v-4), and can my gram run any of it? See /inferences/minicpm-v-4

open a gramx with this ask →epn.family.minicpm-v-4