[whatmodel.sh](../../README.md) / MiniCPM-V 2.6
# MiniCPM-V 2.6
OpenBMB · released 2024-08 · 8.1B params · minicpm-model-license
**score 44.1** ±1.8 · #496 of 732 · from 5 benchmark results across 4 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks MiniCPM-V 2.6 among every model on that task or test, worst to best.
task vs the others rank counts tests
vision & documents ▁●▂▂▁▃▄▂▆▇█▄ #147 of 153 99% 5
vision & documents · #147 of 153 · 99% of its score
Beats [Gemma 3 4B](gemma-3-4b.md) here, though they rank higher overall. Trails [Qwen3.5 0.8B](qwen3-5-0-8b.md), [North Micro Vision Instruct (2.4B)](north-micro-vision-instruct.md), [Molmo 7B-D (0924)](molmo-7b-d.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
CharXiv descriptive questions 57.0% ▄●▄▄▄·██·▄·█ beats 2/12 easy · avg 69% 43% at its level
Arena vision (style controlled) 962.4 ▁▁●▂▂▂▃▄▅▅█▇ beats 3/152 avg 1061 22% blunt
CharXiv (reasoning questions) 31.0% ▂▂▂●▃▂·▂▂··▂ beats 3/13 hard · avg 34% 16% typical
CapArena-Auto detailed image capt… -38.0000 ▃▃●·▃█·▅▃▃▃▃ beats 2/12 avg -4.498 14% typical
PhysBench physical-world understa… 44.9% ▅▃█··▅●····▃ beats 8/9 hard · avg 42% 3.5% blunt
source results share of the score
CharXiv leaderboard 2 59% ▇▇▇▇▇▇▇
LMArena leaderboard dataset 1 22% ▇▇▇
CapArena: Benchmarking and Analyzing De… 1 14% ▇▇
PhysBench: Benchmarking and Enhancing V… 1 3% ▇
prior: pull toward the middle 1% ▇
weighted average = 44.1
Licences: CharXiv leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · CapArena: Benchmarking and Analyzing Detailed Image Captioning in the LLM Era ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · PhysBench: Benchmarking and Enhancing Vision-Language Models for Physical World Understanding ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run MiniCPM-V 2.6 (Q4_K_M, needs ~4.7 GB, from [huggingface.co/openbmb](https://huggingface.co/openbmb/MiniCPM-V-2_6-gguf/tree/48fe6436abf57b3df6ec34f73cdc1fb4b740acb0))
```sh
llama cli -hf openbmb/MiniCPM-V-2_6-gguf:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf openbmb/MiniCPM-V-2_6-gguf:Q4_K_M
```
Other versions: Q8_0 (~8.1 GB) `llama cli -hf openbmb/MiniCPM-V-2_6-gguf:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run MiniCPM-V 2.6 (Q4_K_M, needs ~4.7 GB, from [huggingface.co/openbmb](https://huggingface.co/openbmb/MiniCPM-V-2_6-gguf/tree/48fe6436abf57b3df6ec34f73cdc1fb4b740acb0))
```sh
lms get https://huggingface.co/openbmb/MiniCPM-V-2_6-gguf@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~8.1 GB) `lms get https://huggingface.co/openbmb/MiniCPM-V-2_6-gguf@Q8_0 -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run MiniCPM-V 2.6 (Q4_K_M, needs ~5.36 GB, from [ollama.com](https://ollama.com/library/minicpm-v/tags))
```sh
ollama run minicpm-v:8b-2.6-q4_K_M
```
Other versions: Q8_0 (~10.22 GB) `ollama run minicpm-v:8b-2.6-q8_0`
Model repository and documentation: [huggingface.co/openbmb](https://huggingface.co/openbmb/MiniCPM-V-2_6)
[interactive version](https://whatmodel.sh/model/minicpm-v-2-6) · [JSON](https://whatmodel.sh/api?format=json&all=1)