☁️BizyAir Veo3.1 Text To Video

The simplest of the three Veo nodes in this pack: no image inputs at all, just a text prompt and a model choice. If you don't have a starting frame, or you want Veo to invent the whole scene from scratch, this is the one. The other two nodes in the family - plain image-to-video and the reference-images variant - both exist because someone already had visual material to anchor the generation to. This node assumes you don't.

What it does

Calls Google's Veo 3.1 through your BizyAir account and returns a rendered clip from a text prompt alone. There's genuinely no open-source model that matches Veo on this - the open ecosystem's strongest local option (Wan 2.2) has no native audio at any version, while Veo's whole differentiator is generating synchronized audio and dialogue alongside the video. That gap is real and it's why people reach for a paid API node like this one instead of staying fully local.

The tradeoff is the usual one for anything running on someone else's servers: you're spending BizyAir credits per clip, you don't control the content filter, and there's a network round trip involved instead of a GPU spinning up locally.

Inputs and outputs that matter

• prompt - multiline text describing the whole scene: subject, action, setting, camera behavior. Since there's no image to anchor composition, the prompt is doing all the work here - be more descriptive than you would for an i2v generation where the starting frame already establishes half the scene.
• model - veo3.1-fast for iterating, veo3.1-pro when you're happy with a prompt and want the better render. Same tradeoff as the i2v version: pro costs more and takes longer.
• aspect_ratio (optional) - 16:9 or 9:16.

Output is video, plus bizyair_model_name, which currently reports "veo3.1" for both model tiers - it identifies the family, not which specific tier ran, so don't build logic around it.

How to install it

Same as the rest of the BizyAir pack:

• ComfyUI Manager - search "BizyAir", click Install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/siliconflow/BizyAir.git, restart.
• Comfy CLI - comfy node install bizyair.

Set up your API key before your first run - the in-app "click to login" prompt handles it, or register directly at bizyair.siliconflow.cn. Without a key, none of the API nodes in this pack do anything.

Common issues & troubleshooting

A prompt that's fine for local generation gets rejected here. You're not talking to an uncensored local checkpoint - this is Google's content filter, applied on their end, and there's no local override. If a generation bounces with no useful error, assume it's a policy hit before assuming it's a bug.

Motion doesn't match what you described. Text-to-video with no image anchor gives the model much more freedom to interpret loosely than an image-conditioned generation does. If you're getting compositions you don't want, describing camera framing explicitly (wide shot, close-up, static camera) tends to help more than adding adjectives.

BizyAir installed itself and you're wary of it. A known bug in older ComfyUI Manager versions occasionally auto-installed BizyAir while trying to resolve a missing node from an unrelated workflow, which is where the "is this malware" threads online come from. It isn't - it's a real commercial node pack from SiliconFlow - but if you don't want a metered cloud dependency in your setup, it's a normal uninstall from the Manager, nothing special required.

Nothing happens on Queue. Check your API key and BizyAir credit balance before anything else. There's no local fallback path in this node - it either reaches the cloud or it doesn't run.