The Problem
DeepSeek is smart, fast, and cheap. But it can't see images.
Every time I wanted to ask about a photo, screenshot, or diagram, I had to switch to another AI. It was annoying.
So I built a fix.
What It Does
Upload any image + ask a question. The tool uses GLM-4V for vision recognition, then passes the description to DeepSeek for reasoning. You get DeepSeek's intelligence with the ability to see.
It also works without images — just ask questions like normal DeepSeek.
Features
- 📷 Image recognition — upload photos, screenshots, diagrams
- 💬 Text chat — same DeepSeek you know
- 🌐 Web search toggle — flip it on for real-time info
- 🆓 100% free — no signup, no limits
Try It
Built in a few hours. Uses DeepSeek API + GLM-4V. The code is a single Python file. If you want to self-host, DM me.
Feedback welcome. What should I add next?
Top comments (0)