GitHub

You can see the local AI model that you can run on your PC for free.LLM Checker

"LLM Checker" is a CLI tool that scans your PC's hardware and recommends locally executable LLM. It features a fully integrated connection with Ollama.

Pavelevich/llm-checker: LM LM

https://github.com/Pavelevich/llm-checker

◆Feature

PetsGitHubThe characteristics of LLM Checker are as follows:

・Synced Ollama SQLite catalog is bundled with more than 200 and can be updated from Ollama if necessary

・4D scoring engine enables proper weight of quality, speed, conformity and context depending on the use case

Hardware detection:GPU(Apple Silicon・NVIDIA CUDA・AMD ROCm・Intel Arc) andCPU・RAM acceleration backend

・Predict memory usage compared to actual Ollama size bytes per parameter

・Node.js Pure that can work after 16English・Live display of AI execution speed per second with model output

In addition, there is a llmfit in the same tool, but there is a difference between directions.

| Tools | Main Focus | Typical Output |

| LLM Checker | Hardware-enabled model selection for local reasoning | Ollama pull/run command |

| llmfit | LLM workflow support and model calibration evaluation | Various optimization workflow selection heuristic |

According toGitHubGitHub’s view, it is recommended to use LLM Checker first if you want to get the goal of what to do with this machine. If you want to experiment more easily, you can expect synergistic effects by using both tools.

Contact

This timeWindows Run LLM Checker on PC WSL (Ubuntu). LLM Checker depends on Node.js, so you need to install Node.js (v16 or later) andpmpm in advance. To check the version of Node.js orpmpm installed, run the following command in the terminal:

# Check Node.js version:

node

Check #のバージョンpm version:

npm -v

If the version is not displayed in the above command, you need to install it. The curl command is required for installation.

sudo apt update

sudo apt install

Next, execute the following command to install the latest Node.js/npm at the time of writing the article.

Download and install #するvm

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.4/install.sh

\. "$HOME/.nvm/nvm.sh"

# Download and install Node.js

nvm install 24

The LLM Checker installation itself executes the following commands:

pm install

You need to adjust each PC at the time of initial execution after installation. First, execute the following command to detect PC hardware (CPU, GPU, RAM, acceleration backend):

llm-checker hw-detect

The result is as follows:

Next, execute the following command and check the recommended models by category (such as coding, reasoning, and multimodal).

llm-checker recommend

Various recommended models and pull commands are included.

Next, execute the following command to update the local SQLite model catalog with the latest Ollama reference.

llm-checker

Updated.

Now you have completed the adjustment, you can use the auto-モデルed model and metrics to execute the instructions to AI.

llm-checker ai-run Python"

oution results were noted as "ollama is not installed" as follows.

Install zstd when installing ollama.

sudo apt install

curl -fsSL https://ollama.com/install.sh

After the installation, I added instructions to AI again, and now I noticed that there is no model.

Install "llama2:7b" with the pull command after some of the first models to be installed.

ollama pull llama2:7b

When I asked AI again, I had a strange output. If you read the output contents, "codellama is not installed".

Run the pull command to install codellama.

ollama pull

The Python code that outputs "hello world" was output. The value of "Confidence" in the output result is also 88% in llama2:7b, but it is 100% in codellama.

If you execute "ai-run" after the second time, you need to adjust again because it is already adjusted, and you can add "--calibrated" to the option.

llm-checker ai-run

Contact

If you are using Ollama, it is a very strong tool that lets you define the appropriate pull command according to the situation and check the actual operation. If you are interested in running local AI on your PC, please use LLM Checker.

・Re ・ articles

"AgentStop" is developed by Brave, a technology that solves "the problem that local AI consumes battery and GPU resources while moving with Zulzul" - GIGAZINE

Free "ComfyUI" and "Open WebUI" to move local AI model with GPUDockerPuget Systems Docker App Packs

"Lemonade" which can easily introduce local AI environment for free, Windows ·Linux・macOSAMD GPUNPUGIGAZINE

"mesh-llm" that can collect resources from multiple PCs and run a huge AI model locally - GIGAZINE

llmfit - GIGAZINE

Windows・macOS・LinuxAndroid・iOSOpenClaw - GIGAZINE