Google Veo3.1 Fast Text To Video Fast (Official Edition)

Here's where the Veo family starts to look like a real video tool instead of a three-input toy. The Official Edition of Veo 3.1 Fast text-to-video keeps the Fast tier's speed, but adds the controls the Channel Edition stripped out: duration, generate_audio, a negative_prompt, and a seed. Same cloud model, same pack, but this node actually lets you steer.

The "Official Edition" name is the key. Where the Channel Edition routes through BizyAir's simplified surface, the Official Edition mirrors what the provider's own API exposes - which is why the parameter set suddenly matches what you'd expect from a real video generation endpoint. It's the version you reach for when iteration matters, because seed and negative_prompt are the two things that make reproducible, steerable iteration possible at all.

Why you'd reach for it

Because "give me video" isn't the same as "give me this specific video." With seed you can lock or vary a result; with negative_prompt you can push back against the artifacts you keep seeing - weird hands, flicker, unwanted camera drift. generate_audio defaults to true, which is the right default: native audio is Veo's genuine capability gap over every open model, and turning it off should be a deliberate choice, not an accident. duration defaults to 4. These four inputs turn a one-shot API call into something you can actually direct.

How it works

The required set is prompt, resolution (720p default), duration (default 4), and generate_audio (default true). Optional: negative_prompt, aspect_ratio (16:9 default), and seed (a standard int range up to 2^31−1). The node submits through BizyAir to Veo 3.1 Fast, polls, and returns video (VIDEO - save or preview like any local render), request_id (STRING - trace ID), and urls (STRING - clip location).

Two optional workflow inputs round it out: api_config (BIZYAIR_OPENAPI_CONFIG) overrides base_url and api_key per request, and skip_error makes a failed call output a placeholder instead of aborting the queue.

Install and setup

The same one-time install as every node in this pack:

cd ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt

Restart ComfyUI (or install "BizyAirPlus" via ComfyUI Manager). No model downloads - the bizyair-cloudberry and bizytrd pip packages auto-install on startup. Click the BizyAirPlus button in the action bar to ON, create an account at bizyair.ai, enter the API key when prompted (or in Settings > BizyAirPlus > API Key), and queue.

Common issues

• Reproducibility - seed at 0 is "use a random seed" territory for most APIs; set an explicit value when you want to re-run the same shot. The author's own schema default is 0, so don't assume repeated queues reproduce anything until you've pinned it.
• Artifacts persist - that's what the negative_prompt is for; feed it the specific failure you keep seeing rather than a generic tag dump.
• Audio you didn't want - generate_audio defaults on. Flip it off for silent placeholder clips if you're compositing your own sound.

Pack caveat: BizyAir auto-installs and sends your prompts to the cloud. Legitimate SiliconFlow product, but keep the switch OFF for local-only sessions.