Vllama

Vllama 🦙

One CLI for everything AI — locally or on free cloud GPUs.

Image generation · AutoML · Local LLMs · Speech · Object Detection · 3D · VS Code

pip install vllama

Get Started View on GitHub

---

What is Vllama?

Vllama is a single CLI tool that puts state-of-the-art AI at your fingertips — without needing a powerful GPU or writing any code.

No GPU? No problem.

Vllama can offload heavy models like Stable Diffusion to Kaggle's free GPU and automatically download the results to your machine. All you need is a Kaggle account.

---

Features

🎨

Image Generation

Run Stable Diffusion and other diffusion models locally or on Kaggle GPUs.

🤖

Local LLMs

Spin up any HuggingFace chat model as a local REST API server in one command.

🆚

VS Code Extension

Chat with your locally running LLM directly inside VS Code's native chat panel — zero API keys.

📷

Object Detection

Run YOLO on images and videos from the terminal instantly.

🎬

Video Generation

Generate videos from text prompts using text-to-video models.

🔊

Speech

Text-to-speech and speech-to-text using local models — no cloud API needed.

🖼️

Image/Video to 3D

Generate 3D .ply models from images or videos via Kaggle GPU.

🏆

AutoML

Preprocess any CSV and auto-train 9+ ML models with hyperparameter tuning in two commands.

---

5-Minute Examples

Image GenerationAutoMLLocal LLMObject Detection

# Login to Kaggle once (free account)
vllama login --service kaggle --username YOU --key YOUR_KEY

# Generate an image on free Kaggle GPU
vllama run stabilityai/sd-turbo --service kaggle --prompt "A cyberpunk city at night"

# Step 1: Preprocess your CSV
vllama data --path housing.csv --target price

# Step 2: Train 9 models and get a leaderboard
vllama train --path ./output_folder_YYYYMMDD_HHMMSS --target price

# Terminal 1: Start local LLM server
vllama run_llm Qwen/Qwen2.5-Coder-0.5B-Instruct

# Terminal 2: Chat with it
vllama chat_llm

# Detect objects in a photo
vllama detect_image --path photo.jpg

# Detect objects in a video
vllama detect_video --path video.mp4

---

Installation

pip install vllama

That's it. See Installation for environment-specific setup and optional dependency groups.

---

Next Steps

📦 Installation →

Full install guide including optional dependencies and environment setup.

🚀 Quickstart →

Five hands-on examples to get you running in minutes.

☁️ No GPU Guide →

Run heavy models for free using Kaggle's GPU — Vllama's standout feature.

📚 Command Reference →

Every command, every flag, every output explained.