SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2408.01800 · 2024

MiniCPM-V: A GPT-4V Level MLLM on Your Phone

arXiv · PDF · Open in the Atlas

Code that ran

11 of 14 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 8 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2408.01800.json · badge: /api/ran/2408.01800.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
OpenBMB/MiniCPM-o canonical (classifier label)
classifier basis: census
9 ran · 3 no recorded run
openbmb/minicpm-v canonical (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
FunctionStatusWhere it lives
collect_chunks Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
OpenBMB/MiniCPM-o/web_demos/minicpm-o_2.6/vad_utils.py
code served (permissive licence) · get_code("0f35574f9b09041b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (6,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a sample audio array (float32, small size)
audio = np.array([0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0], dtype=np.float32)

# Create chunks list with dict entries containing 'start' and 'end' keys
chunks = [
    {'start': 0, 'end': 3},
    {'start': 5, 'end': 8}
]

ARGS = {
    'audio': audio,
    'chunks': chunks
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (6,) — shape (6,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
conversation_to_ids_minicpm Ran
ran; our contract draft was wrong, not the code
OpenBMB/MiniCPM-o/finetune/dataset.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from 1xbq1/fedmllm, is pointer only (licence: NONE) · get_code("d6c1484ab50b4e26")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from 1xbq1/fedmllm, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

data_collator Ran
ran; our contract draft was wrong, not the code
OpenBMB/MiniCPM-o/finetune/dataset.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from 1xbq1/fedmllm, is pointer only (licence: NONE) · get_code("4212780bf7da9ddd")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from 1xbq1/fedmllm, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

expand_question_into_multimodal Ran
ran; our contract draft was wrong, not the code
OpenBMB/MiniCPM-o/chat.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from rlhf-v/rlaif-v, is pointer only (licence: NONE) · get_code("f067ba49686d26ba")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from rlhf-v/rlaif-v, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_2d_sincos_pos_embed Ran
ran, honoured the contract we drafted
OpenBMB/MiniCPM-o/omnilmm/model/resampler.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from lluosi/RDB-CL, is pointer only (licence: NONE) · get_code("77e8a3ac46f3afec")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from lluosi/RDB-CL, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_abs_pos Ran
ran; our contract draft was wrong, not the code
OpenBMB/MiniCPM-o/omnilmm/model/resampler.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from kesenzhao/uv-cot, is pointer only (licence: NONE) · get_code("50c4398a8645438a")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from kesenzhao/uv-cot, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_file_extension Ran
ran; our contract draft was wrong, not the code
OpenBMB/MiniCPM-o/web_demos/minicpm-o_2.6/chatbot_web_demo_o2.6.py
code served (permissive licence) · get_code("68b3160674a64aba")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os

# The function get_file_extension takes a filename string and returns its extension
# It uses os.path.splitext() which splits the filename into (name, extension) parts
# and returns the extension in lowercase

ARGS = {
    "filename": "example.txt"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_parameter_number Ran
ran; our contract draft was wrong, not the code
openbmb/minicpm-v/finetune/finetune.py
code served (permissive licence) · get_code("1a3baf406bc7de64")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

ARGS = {
    'model': model
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
is_image Ran
ran, violated the contract we drafted
OpenBMB/MiniCPM-o/web_demos/minicpm-o_2.6/chatbot_web_demo_o2.6.py
code served (permissive licence) · get_code("96d7e1c6bf4c29ec")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os

IMAGE_EXTENSIONS = {'.jpg', '.jpeg', '.png', '.bmp', '.tiff', '.webp'}

def get_file_extension(filename):
    return os.path.splitext(filename)[1].lower()

def is_image(filename):
    return get_file_extension(filename) in IMAGE_EXTENSIONS

ARGS = {
    'filename': 'photo.jpg'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
is_video Ran
ran, violated the contract we drafted
OpenBMB/MiniCPM-o/web_demos/minicpm-o_2.6/chatbot_web_demo_o2.6.py
code served (permissive licence) · get_code("fa9adb328f246208")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os

VIDEO_EXTENSIONS = {'.mp4', '.mkv', '.mov', '.avi', '.flv', '.wmv', '.webm', '.m4v'}

def get_file_extension(filename):
    return os.path.splitext(filename)[1].lower()

def is_video(filename):
    return get_file_extension(filename) in VIDEO_EXTENSIONS

# Setup code
ARGS = {
    'filename': 'sample_video.mp4'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
upload_img Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
openbmb/minicpm-v/web_demos/web_demo.py
code served (permissive licence) · get_code("2a1ffa5243150412")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
from PIL import Image

# Create a small numpy array that can be converted to PIL Image
# Image.fromarray expects uint8 array, so create a small RGB image (4x3x3)
image_array = np.random.randint(0, 256, size=(4, 3, 3), dtype=np.uint8)

# Create a chatbot list (starts empty, will have tuples appended)
chatbot = []

# Create an app_session dict with the expected keys
app_session = {
    'sts': 'some_status',
    'ctx': ['some', 'context'],
    'img': None
}

ARGS = {
    'image': image_array,
    '_chatbot': chatbot,
    '_app_session': app_session
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
conversation_to_ids Not yet run
unverified (harvested, no recorded run)
OpenBMB/MiniCPM-o/finetune/dataset.py
code served (permissive licence) · get_code("a32cd6f57d6772ae")
How it ran

no invocation recorded.

get_2d_sincos_pos_embed_from_grid Not yet run
unverified (harvested, no recorded run)
OpenBMB/MiniCPM-o/omnilmm/model/resampler.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from Zhangwenyao1/DreamVLA, is pointer only (licence: NONE) · get_code("2884deadbb637c3e")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from Zhangwenyao1/DreamVLA, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

get_speech_timestamps Not yet run
unverified (harvested, no recorded run)
OpenBMB/MiniCPM-o/web_demos/minicpm-o_2.6/vad_utils.py
code served (permissive licence) · get_code("a3a415371d24af80")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 11 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 4 the same month, 8 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
get_2d_sincos_pos_embed_from_grid 190 other papers
104 dated earlier, 6 the same month, 79 later, 1 same year, order not determinable · 65 only as an assembly brick
1807.10221 (2018-07)
Unified Perceptual Parsing for Scene Understanding
2105.12723 (2021-05)
Nested Hierarchical Transformer: Towards Accurate, Data-Efficient and Interpretable Visual Understanding
2111.06377 (2021-11)
Masked Autoencoders Are Scalable Vision Learners
and 187 more not named here
get_2d_sincos_pos_embed 95 other papers
62 dated earlier, 3 the same month, 29 later, 1 same year, order not determinable · 33 only as an assembly brick
1807.10221 (2018-07)
Unified Perceptual Parsing for Scene Understanding
2105.12723 (2021-05)
Nested Hierarchical Transformer: Towards Accurate, Data-Efficient and Interpretable Visual Understanding
2111.06377 (2021-11)
Masked Autoencoders Are Scalable Vision Learners
and 92 more not named here
get_abs_pos 8 other papers
5 dated earlier, 3 later
2311.16922 (2023-11)
Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding
2402.13561 (2024-02)
Cognitive Visual-Language Mapper: Advancing Multimodal Comprehension with Enhanced Visual Knowledge Alignment
2405.17220 (2024-05)
RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness
and 5 more not named here
collect_chunks 2 other papers
1 the same month, 1 later
2408.16725 (2024-08)
Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming
2410.11190 (2024-10)
Mini-Omni2: Towards Open-source GPT-4o with Vision, Speech and Duplex Capabilities
conversation_to_ids_minicpm 2 other papers
1 dated earlier, 1 later
2405.17220 (2024-05)
RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness
2411.14717 (2024-11)
FedMLLM: Federated Fine-tuning MLLM on Multimodal Heterogeneity Data
data_collator 2 other papers
1 dated earlier, 1 later
2405.17220 (2024-05)
RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness
2411.14717 (2024-11)
FedMLLM: Federated Fine-tuning MLLM on Multimodal Heterogeneity Data
get_parameter_number 2 other papers
1 dated earlier, 1 later
2405.17220 (2024-05)
RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness
2411.14717 (2024-11)
FedMLLM: Federated Fine-tuning MLLM on Multimodal Heterogeneity Data
get_speech_timestamps 2 other papers
1 the same month, 1 later
2408.16725 (2024-08)
Mini-Omni: Language Models Can Hear, Talk While Thinking in Streaming
2410.11190 (2024-10)
Mini-Omni2: Towards Open-source GPT-4o with Vision, Speech and Duplex Capabilities
conversation_to_ids 1 other paper
1 dated earlier
2405.17220 (2024-05)
RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness
expand_question_into_multimodal 1 other paper
1 dated earlier
2405.17220 (2024-05)
RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness
upload_img 1 other paper
1 dated earlier
2405.17220 (2024-05)
RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The recent surge of Multimodal Large Language Models (MLLMs) has fundamentally reshaped the landscape of AI research and industry, shedding light on a promising path toward the next AI milestone. However, significant challenges remain preventing MLLMs from being practical in real-world applications. The most notable challenge comes from the huge cost of running an MLLM with a massive number of parameters and extensive computation. As a result, most MLLMs need to be deployed on high-performing cloud servers, which greatly limits their application scopes such as mobile, offline, energy-sensitive, and privacy-protective scenarios. In this work, we present MiniCPM-V, a series of efficient MLLMs deployable on end-side devices. By integrating the latest MLLM techniques in architecture, pretraining and alignment, the latest MiniCPM-Llama3-V 2.5 has several notable features: (1) Strong performance, outperforming GPT-4V-1106, Gemini Pro and Claude 3 on OpenCompass, a comprehensive evaluation over 11 popular benchmarks, (2) strong OCR capability and 1.8M pixel high-resolution image perception at any aspect ratio, (3) trustworthy behavior with low hallucination rates, (4) multilingual support for 30+ languages, and (5) efficient deployment on mobile phones. More importantly, MiniCPM-V can be viewed as a representative example of a promising trend: The model sizes for achieving usable (e.g., GPT-4V) level performance are rapidly decreasing, along with the fast growth of end-side computation capacity. This jointly shows that GPT-4V level MLLMs deployed on end devices are becoming increasingly possible, unlocking a wider spectrum of real-world AI applications in the near future.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2408.01800")
get_code_for_paper("2408.01800")
have("2408.01800")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2408.01800.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2408.01800.svg)](https://syntology.ai/paper/2408.01800)

Connect an agent — have() is free.