MiniMax Text to Video

This is the node you reach for when you want the simplest possible demonstration of what ComfyUI's built-in video nodes are: prompt in, video out, nothing in between. MiniMax's T2V-01 text-to-video model, wrapped in a single node with three widgets and one flat price. No reference images, no end frames, no storyboards, no aspect ratio picker - just your prompt, a model choice, and a seed. If you're new to this whole partner/video category and want to understand the model without reading a manual, this is the one to drag onto the canvas first.

It's an API node, exactly like the rest of the family: ships with ComfyUI core, sends your prompt to MiniMax's API through Comfy Org's backend, and bills your Comfy account in credits. The flat price badge (roughly $0.43 a video) doesn't change with your prompt or duration, which makes it the most predictable node in the entire category for budgeting.

The inputs that matter

• prompt_text - the video description. This is 100% of the creative work. MiniMax's T2V responds well to concrete, visual, motion-inclusive language: subject, action, camera, mood. A paragraph of "beautiful, cinematic" adjectives gets you less than two sentences of actual scene description.
• model - T2V-01 (default) or T2V-01-Director. The Director variant is MiniMax's tier for more deliberate, controlled generation - the one to try if plain T2V-01 feels like it's ignoring your camera direction.
• seed - an optional 64-bit seed. Family standard: it's a "force a re-run for a different take" control, and results stay non-deterministic no matter what you set.

Output is a single VIDEO object - wire it into a save or preview node, or let the UI display it.

What this node is for

Honestly? Testing, learning, and quick-turnaround drafts. The whole point of the minimal surface is that there's nothing to misconfigure. Where people get burned is almost never the node itself - it's the expectations around it:

• It's closed and paid. MiniMax sits in the "all closed source" column of every community roundup. No weights, no local fallback, and the flat fee applies to every render, good or bad.
• The "synchronous" label hides a queue. The node submits a task and waits - MiniMax renders, the UI shows a waiting state, and the clip downloads when done. Don't interrupt; the job is already charged.
• Prompt is everything, so iterate on the text. Since you can't tune resolution or duration to save money, the only free lever you have is writing a better prompt. Use it.

If you're curious whether the built-in API video nodes are worth using at all, this is the cheapest, least intimidating way to find out. One prompt, one wait, one flat bill - and you'll know exactly what you think of the whole category.