Open XinYu chat on the canvas and the default brain is DeepSeek V4.1 Flash 1017 — free users can use it too.
What it does
- Thinks before it acts: it works out a plan first, then builds nodes, wires them up and writes prompts on your canvas.
- Reads your images: hand it a reference image and it understands the picture before it starts.
- Available in text nodes and storyboards too: you can give it images; for videos, switch to Gemini 2.5 Flash — if you pick the wrong one, you'll be told before anything is charged.
Why "1017"
The model picker now shows "DeepSeek V4.1 Flash 1017", so it's easy to tell apart from the earlier V4.1 Flash. Chats and nodes that already use it carry over automatically.
Every model's price is shown in the picker — take a look before you switch.