GitHub - XInTheDark/raycast-g4f: Raycast extension to use GPT, Claude, Llama, and more... all for FREE! + Full support for custom APIs.
Summary: This open-source Raycast extension enables users to access powerful AI models like GPT-4, Claude, Llama, and more—completely free—via custom APIs. It supports streaming responses, chat history, web searches, file uploads, image generation, and custom prompts. With 18+ commands and frequent updates, it's a cost-effective alternative to paid AI tools, prioritizing privacy and flexibility.
1. Overview & Key Features
Raycast-G4F is a free, open-source extension for macOS users that integrates advanced AI models directly into the Raycast app. Key highlights:
- Free access to models like GPT-4, Llama-3, Gemini, and Claude.
- Streaming responses (▶️) for real-time interactions.
- Web search integration (via DuckDuckGo) to fetch the latest info.
- File uploads (text, images, videos, audio) for context-aware AI interactions.
- Custom APIs support for personal or enterprise models.
- Image generation capabilities with state-of-the-art models.
- Chat history storage for seamless conversations.
"If you like the extension, please consider giving it a ✨star✨ tysm!" – the developer, probably
2. Screenshots
3. Getting Started
Prerequisites
- Raycast app (macOS-only; Windows support in progress).
- Node.js v20.18.1+ (older versions may cause issues).
Installation
- Download the latest release or clone the repository.
- Open Terminal in the folder and run:
npm ci --production pip3 install -r requirements.txt # Optional for web search npm run dev - The extension appears in Raycast with all commands.
Updates
- Automatic: Use the "Check for Updates" command (enabled by default).
- Manual: Run
git pull,npm ci --production, andnpm run dev.
4. Providers & Models
The extension supports 20+ models across providers like Nexra, DeepInfra, Blackbox, and Google Gemini. Key examples:
| Provider | Model | Features | Speed | Rating |
|---|---|---|---|---|
| Nexra | gpt-4o | Streaming, web search | Very fast | 8.5/10 |
| DeepInfra | meta-llama-3.3-70b | Streaming, large context | Fast | 8.5/10 |
| Google Gemini | gemini-1.5-pro | Streaming, file upload | Very fast | 9/10 |
| Blackbox | claude-3.5-sonnet | Streaming, coding optimized | Very fast | 8.5/10 |
🔥 Top pick: Google Gemini (free API key required) for balanced speed and accuracy.
5. Features & Customization
Web Search
- Options: Automatic, Balanced, Always (enabled per command/chat).
- Uses DuckDuckGo for privacy-focused results.
Smart Chat Naming
GPT auto-generates chat titles (e.g., "Python Debugging Session") for easy organization.
Persistent Storage
Backs up chat data and custom commands to files (disable for sensitive info).
Code Interpreter (BETA)
Runs Python code locally (supported by select DeepInfra models).
Custom APIs
Connect any OpenAI-compatible API via the help guide.
6. FAQs
Why isn't it on the Raycast store?
- Third-party providers can't be fully vetted by Raycast.
- Faster updates via GitHub.
How does it compare to paid Raycast AI?
- Freedom: Use any API (OpenAI, Gemini, etc.).
- Cost: Free with built-in providers; pay only for custom APIs.
- Privacy: Open-source, data stored locally.
Troubleshooting
- No response? Try another provider or check GitHub issues.
- Slow? Switch to a "Fast" or "Extremely fast" model.
7. License & Acknowledgements
- License: GPLv3 (see LICENSE.txt).
- Inspired by: gpt4free (Python).
- Libraries: g4f-image, gemini-ai-sdk.
Conclusion
Raycast-G4F is a powerhouse for AI enthusiasts seeking free, flexible access to cutting-edge models. With streaming, custom APIs, and robust features, it's a must-try for users tired of paid subscriptions. Start exploring today—your AI workflow will thank you! 🚀
Need a summary like this?
Get instant summaries with Harvest