The Canvas LLM plugin offers a visual, node-based interface for interacting with large language models. Instead of the usual linear chat, it lets users create branching conversations using canvas-style cards. You input prompts in `Input` cards and receive model outputs in `Generate` cards, allowing for multiple lines of thought or variations in a single workspace. Each block can independently connect to a different model, making it flexible for experimentation or comparing responses. The plugin supports eight API providers, including [Google](https://ai.google.dev/gemini-api/docs/pricing), [OpenRouter](https://openrouter.ai/models?max_price=0), and [SambaNova](https://cloud.sambanova.ai/plans/pricing), but users must bring their own API keys. This makes it ideal for research-heavy workflows or creative projects that benefit from nonlinear thinking.