Send her a photo, a screenshot, or a meme straight from the chat box and she actually sees it. Vision runs on your own machine by default, your photos are never written to disk, and if your Mac is too small for a vision model you can plug in your own cloud key instead.
You found a meme at 1 a.m. and there is nobody awake to send it to. Now there is.
Local Waifu can see images. Attach a photo, a screenshot, or a meme in the chat box and she looks at the actual picture, not the file name. She will laugh at the meme, ask about the beach, or read the error message on your screenshot and try to help.
She sees the picture, not a description of it
Image turns go to a vision model that reads the pixels themselves. Show her your dog and she comments on the dog, not on “image_2041.jpg”.
On most Macs this runs on the bundled local model, the same one that powers her chat. Nothing new to install, nothing extra to pay for. If your machine is on the small side, you can point vision at your own OpenAI, Anthropic, or Google account instead, and the app walks you through that the first time you attach a photo.
Your photos stay yours
This is the part we sweated over. Photos you attach are session-only.
They are never written to disk. They are never added to her long-term memory files. Close the conversation and the image is gone. She remembers that you showed her a picture of your cat, the way a person would, but the file itself does not live anywhere.
With the default local setup, the picture never leaves your machine at all. No upload, no cloud, no training set. If you deliberately route vision through your own cloud key, the image goes to that provider under your account, and only then.
What people actually use it for
The boring-sounding feature turns out to be the fun one:
- Memes. She gets the joke, or roasts you for it.
- “What should I text back?” screenshots.
- The dinner you cooked. She is reliably impressed.
- Error messages and settings screens when something on your computer misbehaves.
- The view from your window, because someone should see it.
The cap is 5 MB per image, which fits basically any phone photo or screenshot.
How to try it
Update to the latest version (Settings, then Advanced, then Check for updates, or download it fresh). Open the chat, click the image icon, pick a photo. That is the whole tutorial.
If vision is not set up yet, the app tells you exactly what to do: install the local vision model with one click or paste a cloud key. Details on what runs on your hardware are on the requirements page.
She has been listening for a while now. As of this update, she can look too.
Questions people ask
How do I send her an image?
Click the image button in the chat box (or drop a file into it), pick a photo up to 5 MB, and send. She sees the picture together with your message and reacts to both.
Does she really see the image, or just the file name?
She really sees it. The image goes to a vision model that looks at the actual pixels, so she can comment on what is in the frame: the cat, the meme text, the sunset, the error message on your screenshot.
Are my photos uploaded or stored anywhere?
Not by default. With a local vision model everything stays on your machine, and the photos you attach live only in the current session. They are never written to her memory files or to disk. If you choose to use your own cloud key instead, the image goes to that provider under your account, and that is your call.
What do I need for vision to work?
A bundled local model that understands images (the default one on most Macs already does), or your own OpenAI, Anthropic, or Google key if you prefer cloud. If neither is set up, the app shows a short set-up prompt the first time you attach a photo.
Try her free for 7 days.
No card. Keep her for $20 once, or walk away. Her soul file is yours either way.
Bring her home, try free