Molmo 7B vs MiniCPM-V 4.5 for images
Two small vision models that describe and read images on an 8 to 16 GB machine
Facts from the reviewed catalog
| Specification | Molmo-7B-D-0924 | MiniCPM-V-4_5 |
|---|---|---|
| Task | Vision | Vision |
| Parameters | 8B | 8.7B |
| License | Apache 2.0 | Apache 2.0 |
| Commercial use | Allowedlicense text | Allowedlicense text |
| Quality · LMArena 2026-10-19 · vision | 994 ± 14980–1,008 · 2,815 votes · rank 158 · source | Not on the LMArena vision leaderboard of 2026-10-19 · source |
| Memory | Q4 planning estimate (~6 GB of weights before the machine's reserve) | Q4 planning estimate (~7 GB of weights before the machine's reserve) |
| Checkpoint | official card | official card |
Only Molmo-7B-D-0924 has a result in LMArena 2026-10-19 · vision (994 ± 14); MiniCPM-V-4_5 has none, so the two are not ranked from this source and no score is inferred for it. Blind pairwise votes by people on hosted outputs, CC BY 4.0. A rating is relative; overlapping intervals mean the order is not settled. Scale: Arena rating ± half the confidence interval.
Memory on three machines
Each machine's own reserve is added (graphics cards 2 GB; unified and system memory 25%, 2 to 8 GB). Planning estimates, not measured runs.
| Machine | Molmo-7B-D-0924 | MiniCPM-V-4_5 |
|---|---|---|
| RTX 4060 · 8 GB VRAM | ~8 GBFast memory holds it | ~9 GBIn memory, outside the fast part · offload to PC RAM (32 GB assumed) |
| RTX 3060 · 12 GB VRAM | ~8 GBFast memory holds it | ~9 GBFast memory holds it |
| MacBook Air 13-inch M4 · 16 GB unified | ~10 GBFast memory holds it | ~11 GBFast memory holds it |
Molmo-7B-D-0924: Q4 planning estimate (~6 GB of weights before the machine's reserve). MiniCPM-V-4_5: Q4 planning estimate (~7 GB of weights before the machine's reserve). Formulas
Everything that runs on each machine: RTX 4060 · 8 GB VRAM · RTX 3060 · 12 GB VRAM · MacBook Air 13-inch M4 · 16 GB unified
Which to choose
Both are about 8 billion parameters and Apache 2.0, so they need about 6 and 7 GB of Q4 weights before the image tokens; on an 8 GB card the table keeps Molmo in the fast memory and MiniCPM-V just outside it. Molmo 7B-D is Ai2's September 2024 release; MiniCPM-V 4.5 is OpenBMB's newer model. Molmo has a rating in the LMArena vision release; MiniCPM-V 4.5 is not in it, so this source cannot rank them. Check how each runtime handles images before choosing: support for vision models varies more than for text.
Files and run commands
The default file of the most downloaded collected source, or the commands each card documents. Not run by this site.
Molmo-7B-D-0924
Documented ways to run it
Copied as printed in the linked source (read 2026-10-18); not run by this site.
pip install einops torchvisionfrom transformers import AutoModelForCausalLM, AutoProcessor, GenerationConfig
from PIL import Image
import requests
# load the processor
processor = AutoProcessor.from_pretrained(
'allenai/Molmo-7B-D-0924',
trust_remote_code=True,
torch_dtype='auto',
device_map='auto'
)
# load the model
model = AutoModelForCausalLM.from_pretrained(
'allenai/Molmo-7B-D-0924',
trust_remote_code=True,
torch_dtype='auto',
device_map='auto'
)
# process the image and text
inputs = processor.process(
images=[Image.open(requests.get("https://picsum.photos/id/237/536/354", stream=True).raw)],
text="Describe this image."
)
# move inputs to the correct device and make a batch of size 1
inputs = {k: v.to(model.device).unsqueeze(0) for k, v in inputs.items()}
# generate output; maximum 200 new tokens; stop generation when <|endoftext|> is generated
output = model.generate_from_batch(
inputs,
GenerationConfig(max_new_tokens=200, stop_strings="<|endoftext|>"),
tokenizer=processor.tokenizer
)
# only get generated tokens; decode them to text
generated_tokens = output[0,inputs['input_ids'].size(1):]
generated_text = processor.tokenizer.decode(generated_tokens, skip_special_tokens=True)
# print the generated text
print(generated_text)
# >>> This image features an adorable black Labrador puppy, captured from a top-down
# perspective. The puppy is sitting on a wooden deck, which is composed ...MiniCPM-V-4_5
Documented ways to run it
Copied as printed in the linked source (read 2026-10-18); not run by this site.
import torch
from PIL import Image
from transformers import AutoModel, AutoTokenizer
torch.manual_seed(100)
model = AutoModel.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True, # or openbmb/MiniCPM-o-2_6
attn_implementation='sdpa', torch_dtype=torch.bfloat16) # sdpa or flash_attention_2, no eager
model = model.eval().cuda()
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-V-4_5', trust_remote_code=True) # or openbmb/MiniCPM-o-2_6
image = Image.open('./assets/minicpmo2_6/show_demo.jpg').convert('RGB')
enable_thinking=False # If `enable_thinking=True`, the thinking mode is enabled.
stream=True # If `stream=True`, the answer is string
# First round chat
question = "What is the landform in the picture?"
msgs = [{'role': 'user', 'content': [image, question]}]
answer = model.chat(
msgs=msgs,
tokenizer=tokenizer,
enable_thinking=enable_thinking,
stream=True
)
generated_text = ""
for new_text in answer:
generated_text += new_text
print(new_text, flush=True, end='')
# Second round chat, pass history context of multi-turn conversation
msgs.append({"role": "assistant", "content": [generated_text]})
msgs.append({"role": "user", "content": ["What should I pay attention to when traveling here?"]})
answer = model.chat(
msgs=msgs,
tokenizer=tokenizer,
stream=True
)
generated_text = ""
for new_text in answer:
generated_text += new_text
print(new_text, flush=True, end='')Numbers each publisher reports
Each card's own results with its own settings. Publishers choose different tests and settings, so these values are never compared across publishers here.
Molmo-7B-D-0924 · reported by Ai2
MiniCPM-V-4_5 · reported by OpenBMB
Limits and dates
The published quality benchmark does not test a local quantized build. Memory estimates are not speed measurements or guaranteed hardware fits. Repository size, when available, is not VRAM. Formulas
Page updated · Catalog reviewed · LMArena 2026-10-19 · vision