EvoLink Seedance 2.5 Text-to-Video (Official)

Seedance 2.5 is ByteDance's current video generation model, and the headline feature is length: up to 30 seconds per clip, double the 2.0 ceiling, plus a "smart duration" mode where the model decides how long the shot should breathe instead of you guessing. EvoLink Seedance 2.5 Text-to-Video (Official) is that model as a cloud node - closed weights, API-only, the same per-render meter the community has loudly priced out before. You can't run it locally; you can call it.

What it is

Prompt to video, 4 to 30 seconds, 1080p at the top (2.5's quality menu tops out at 1080p - no 4K tier here, which is the one thing the 2.0 node still has over it), with sync audio included at no extra charge. The generate_audio toggle defaults to on, and unlike a lot of video APIs, the tooltip is explicit that it doesn't add to the bill.

Inputs that matter:

• prompt (required) - the shot description, Chinese welcome, ~500 chars suggested.
• duration - 4–30 seconds, default 5, billed by duration.
• auto_duration - the smart knob. Flip it on and duration is ignored; the model picks a length in the 4–30 range and you pay for what it actually outputs. Good for scenes that don't have an obvious length, and it can genuinely surprise you with better pacing.
• quality - 480p / 720p / 1080p. 480p is the cheap lane.
• aspect_ratio - adaptive default, or 16:9, 9:16, 1:1, 4:3, 3:4, 21:9.
• generate_audio - on by default, free.

Outputs: video (connect Save Video or nothing appears), task_id, status_info, result_urls (24-hour expiry), response_json. Default timeout is 30 minutes; a 30-second clip is a long render, so budget your day accordingly.

Install and key

The shared EvoLink install. ComfyUI Manager → search EvoLink → install → full restart (close the console, not just the browser). Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/deeplearning-goethe/comfyui-evolink

Only requests as a dependency; no models, no GPU. Key from evolink.ai/dashboard/keys (sk-), pasted once into api_key - saved locally, blank after, never in a shared workflow JSON.

Is 2.5 worth the upgrade from 2.0?

For longer-form work, yes - 30 seconds is a real scene, not a mood. The smart-duration mode is the feature to actually try first, because it changes the way you prompt: instead of forcing everything to fit 5 seconds, you write the shot and let the model time it. For short, punchy, high-res clips, the 2.0 node's 4K option still has a niche. Either way, the economics are the same wall: this is metered video, the community's pricing objection applies to every Seedance tier, and the discipline that keeps it sane is unchanged - draft cheap, promote winners, don't treat the dropdown as a menu to spend from. Failed and moderation-blocked tasks don't bill, and unchanged params cache without recharging.