Image Generation & Vision

OpenCrabs supports image generation and vision analysis via Google Gemini. These features are independent of the main chat provider — use Claude for chat and Gemini for images.

Setup

1. Get a free API key from aistudio.google.com
2. Run /onboard:image in chat to configure interactively
3. Or add manually:

# keys.toml
[image]
api_key = "AIza..."

# config.toml
[image.generation]
enabled = true
model = "gemini-3.1-flash-image-preview"

[image.vision]
enabled = true
model = "gemini-3.1-flash-image-preview"

Agent Tools

When enabled, two tools become available automatically:

Example prompts:

• “Generate a pixel art crab logo” — agent calls generate_image, returns file path
• “What’s in this image: /tmp/screenshot.png” — agent calls analyze_image via Gemini

Incoming Images

When a user sends an image from any channel, it arrives as <<IMG:/tmp/path>> in the message. The file is already downloaded — the agent can:

• See it directly (if the model supports vision)
• Pass the path to analyze_image for Gemini analysis
• Use the path in bash commands or any tool that accepts file paths
• Reference it in replies with <<IMG:path>> to forward to channels

Model

Both tools use gemini-3.1-flash-image-preview — Gemini’s dedicated image model that supports both vision input and image output in a single request.

Per-Provider Vision Model

Separately from the Gemini analyze_image tool, any provider can have its own vision tool via vision_model. When the user sends an image and the chat model can’t handle it natively, the agent calls the provider’s vision tool — which sends the image to the vision_model on the same provider, gets a text description back, and uses that context to reply.

Example: User sends an image to MiniMax M2.5 (no native vision). The agent calls the vision tool, which sends the image to MiniMax-Text-01, gets the description, and M2.5 replies using that context.

See Provider Setup — Vision Model.