Concurrent Submit | Zhenzhen Google Veo3 Lite

Google's Veo line is the one that made native audio in video a thing - sound effects and dialogue that actually sync with the footage, which open-weights video still hasn't fully matched. The Lite tier is the budget door into that world: veo3.1-lite, cheaper than the flagship models, trading some polish for cost. This node wraps it for ComfyUI via the Comfyui-zhenzhen reseller, and the Concurrent Submit variant is what makes it practical - because the real problem with Veo isn't quality, it's the wait. Lite is faster than Pro, but it's still a video render; queue a batch and let the pool chew through them in parallel.

There's no local model here - it's an API call with the author's key, metered per second of video. Your ComfyUI canvas just becomes the front end.

How the concurrent bit works

The plain Comfly_Googel_Veo3_Lite node blocks until the clip is done. The Submit variant runs it in the pack's shared video pool (10 workers) and returns a task immediately. Collect via Concurrent Collect Videos (10): wire task_1 … task_10, and the collector waits for the batch, returning videos in slot order plus a status string. failure_mode = fail-fast (default) or placeholder.

Ten clips at once is the whole game here. Serially, a batch of Veo jobs is an evening; in the pool, it's one run.

The inputs that matter

• prompt (required) - what happens in the clip.
• model - locked to veo3.1-lite (one choice).
• enhance_prompt - default false. This is the pack's Chinese-to-English translator switch: Veo only really handles English prompts, so if you prompt in Chinese, flip this on.
• aspect_ratio - 16:9 or 9:16. No 1:1 on the Lite tier.
• image1 … image3 - up to three reference images to steer the video (the Lite node does allow image reference).
• enable_upsample - default false. Turns on upscaling for the output; costs more, obviously.
• seed - plain ComfyUI seed, 0–2^31.

Output: one task (COMFLY_VIDEO_FUTURE) → Collector → videos.

Installing it

Ships in Comfyui-zhenzhen:

• ComfyUI Manager → Install Custom Nodes → search "Comfyui-zhenzhen" → install → restart.
• Or cd ComfyUI/custom_nodes && git clone https://github.com/T8mars/Comfyui-zhenzhen and restart.

No model files to fetch. You need a paid token from the author's storefront (ai.t8star.org), in api_key.

Gotchas

The README has a whole section on Veo token groups: the Lite model is only served by specific token groups (the "veo&grok-备用" spare groups), so if you're getting group errors, check which group your token is enrolled in rather than assuming the node is broken. Google's video backend also wobbles - 429/500 = load, retry. And enhance_prompt is the single most-answered question about these nodes: if your Chinese prompt comes back as gibberish, it's because it was never translated.