Images and Vision

Vision-capable models can answer questions about images. 3sparks Chat supports images from the camera, photo library, paste, and PDF page conversion.

Add an image

1. Open a chat with a vision-capable model.
2. Tap + and choose Camera or Photos, or paste an image into the composer.
3. Add a question or instruction.
4. Send the message.

Local and remote vision

• Local Model Vision: supported GGUF models can process images on device when the required projector file is available.
• Remote vision: providers such as OpenAI, Anthropic, Gemini, and OpenRouter can process images when the selected model supports image input.

Image processing

Images may be resized and compressed before sending so they fit provider limits. The app estimates image token usage where possible.

Troubleshooting

• If the model rejects the image, switch to a vision-capable model.
• If image upload is slow, use fewer or smaller images.
• If a local vision model fails, reload the model or confirm the projector file downloaded.
• If the keyboard needs image support, use a reachable provider connection rather than a main-app-only on-device Local Model.