AI Image Renamer

AI Image Renamer CLI is a command-line tool that uses AI vision models to rename image files based on their content, giving your photo collection more descriptive and searchable filenames. It works with Groq's hosted API (fast, free tier) or local LLMs (Ollama, LM Studio, vLLM, and any OpenAI-compatible endpoint) for fully offline renaming.

For Groq, create your free API key here. For full documentation, visit the official docs.

Features

• 🤖 AI: Leverage the latest AI technology to quickly rename your images
• 🏠 Local LLMs: Run fully offline with Ollama, LM Studio, vLLM, or any OpenAI-compatible endpoint
• ⚡️Speed: Groq's fast infrastructure processes your files in milliseconds
• 🔎 SEO: Generated file names are SEO-friendly
• 📚 Batch: Use up to 3 image files within a single command (processed sequentially)
• 👨‍💻 Easy: Renaming files requires only a single command line

Table of Contents

[TOC]

Installation

AI Image Renamer is available through multiple installation methods:

1. Using pipx (recommended)

# Install globally in an isolated environment (requires pipx installed)
pipx install ai-image-renamer

# With local LLM support (Ollama, LM Studio, etc.)
pipx install "ai-image-renamer[local]"

2. Using pip

# Install into the current Python environment
pip install ai-image-renamer

# With local LLM support (Ollama, LM Studio, etc.)
pip install "ai-image-renamer[local]"

3. From the Git repository

# Clone the repository
git clone https://gitlab.com/thaikolja/ai-image-renamer.git
# Navigate into the project directory
cd ai-image-renamer
# Install the package from source
pip install .

4. From a ZIP archive

1. Download the ZIP from the repository.
2. Extract it and run:

# Navigate into the extracted directory
cd ai-image-renamer-main
# Install the package from source
pip install .

5. Run directly from source

# Run via the module entry point without installation
python3 -m ai_image_renamer.cli path/to/image.jpg

After installation, obtain a free Groq API key. The recommended way to configure the tool is via a config.ini file in your working directory. On first run, a commented config.ini is auto-generated — just edit it to set your API key:

# Edit the generated config.ini
GROQ_API_KEY=gsk_your_api_key_here

Alternatively, set it as an environment variable:

export GROQ_API_KEY="your-key-here"

Providers

The tool supports three AI backends, selected via PROVIDER in config.ini or the --provider CLI flag.

Groq (default)

Hosted and fast. Requires a free API key:

PROVIDER=groq
GROQ_API_KEY=gsk_your_api_key_here

Ollama (local)

Fully offline. Your images never leave your machine:

# 1. Install Ollama: https://ollama.com
# 2. Start the server and pull a vision model
ollama serve
ollama pull llava
# 3. Install the CLI with local support
pip install "ai-image-renamer[local]"

PROVIDER=ollama
OLLAMA_HOST=http://localhost:11434/v1
OLLAMA_MODEL=llava:latest

OpenAI-compatible (LM Studio, vLLM, llama.cpp, ...)

Any server speaking the OpenAI Chat Completions API:

PROVIDER=openai
OPENAI_API_BASE=http://localhost:1234/v1
OPENAI_MODEL=my-vision-model

See the documentation for full provider setup and troubleshooting.

Usage

The rename_images command is your entry point to the tool. The AI model, word count, and other options are configurable via CLI flags or config.ini. Some limitations apply:

Basic Usage

To rename a single image:

# Rename a single image file
rename_images path/to/your/image.jpg

To rename multiple images (up to 3 at once):

# Provide up to 3 image paths; append flags after paths
rename_images image1.png image2.jpg path/to/another/image.webp

Use shell glob patterns to select files:

# The shell expands the glob before passing paths to the tool
rename_images ~/Desktop/my-photos/*.png

# Match files containing a keyword in the name
rename_images ~/Photos/bangkok-*.jpg

To rename an image with only 3 words:

# Limit the generated filename to N words
rename_images -w 3 DSC_123.jpg

Override the API key, provider, or model for a single invocation:

# Override API key and model via CLI
rename_images --api-key gsk_xxx --model qwen/qwen3.6-27b

# Use a local Ollama model
rename_images --provider ollama photo.jpg

# Use a custom OpenAI-compatible endpoint
rename_images --provider openai --model my-vision-model photo.jpg

Use glob patterns to select files:

# The shell expands the glob before passing paths to the tool
rename_images ~/Desktop/my-photos/*.png

See rename_images -h for more options, or read the documentation.

Configuration

The tool reads settings from ./config.ini in the current working directory. If the file doesn't exist, it is auto-generated with comments explaining every option:

# AI Image Renamer — Configuration
PROVIDER=groq                          # AI backend: groq, ollama, or openai
GROQ_API_KEY=                          # Your Groq API key (required for groq)
MODEL=qwen/qwen3.6-27b                 # Groq AI vision model (vision required)
OLLAMA_HOST=http://localhost:11434/v1  # Ollama endpoint (for ollama)
OLLAMA_MODEL=llava:latest              # Ollama vision model (for ollama)
OPENAI_API_BASE=                       # OpenAI-compatible endpoint (for openai)
OPENAI_API_KEY=                        # Key for the endpoint (for openai)
OPENAI_MODEL=                          # Model name (for openai)
TEMPERATURE=1.0                        # AI creativity (0.0 – 2.0)
TIMEOUT=30                             # API request timeout in seconds
MAX_RETRIES=3                          # Retries on API failure
DEFAULT_WORD_COUNT=6                   # Default word count (1 – 50)
MAX_FILENAME_LENGTH=100                # Max filename stem length (characters)

[!IMPORTANT]

CLI arguments (e.g. -w) always override the config file values.

Priority: CLI flags → environment variables → config.ini → built-in defaults.

Security

[!WARNING]

Security: Add config.ini to your .gitignore to avoid committing your API key.

Limitations

The hosted Groq provider has the following practical limits to keep in mind (local providers like Ollama are not subject to these):

• Image file size: Each image must be under 20 MB (files larger than that will be rejected by the API).
• Image resolution: Very high-resolution photos (over 33 megapixels, e.g., some DSLR or smartphone camera shots) may need to be resized first.
• Base64 limit: When sending images via API, the base64-encoded data must stay under 4 MB.
• Images per request: The model accepts up to 5 images at once (the CLI caps this at 3 to stay well within bounds).
• Model: The model can be changed via --model CLI flag, GROQ_MODEL environment variable, or MODEL in config.ini. Browse available models at console.groq.com/docs/models.
• Model Reasoning: Reasoning models (Qwen, GPT-OSS) output internal thinking by default. Set REASONING_EFFORT=none in config.ini to suppress it. See Groq reasoning docs for details.

Local LLM requirements

To use local providers, you need the [local] extra installed and a running server:

pip install "ai-image-renamer[local]"

Ollama users also need a vision model pulled (e.g., ollama pull llava) and the server running (ollama serve). Local models are typically slower than Groq but keep all images on your machine.

Contributing

I welcome contributions to AI Image Renamer! Please see the CONTRIBUTING.md file for guidelines on how to contribute.

Author

Kolja Nolte (kolja.nolte@gmail.com)

License

This project is licensed under the MIT License. See the LICENSE file for details. A little self-promotion: If you're not sure which license to use for your project, check out https://whatlicense.org.