Wan2.6 Flash

Wan 2.6 Flash is the API-only follow-up in Alibaba's Wan video line - and that's the whole story of why you'd reach for this node. The open-weights Wan you can run locally stops at 2.2; 2.5, 2.6, and 2.7 shipped without weights, and the community's joke is that Alibaba "can't count to 3" in open releases. Wan 2.6 Flash is what the closed 2.6 tier looks like when you're on the free side of an API: an image-to-video model that animates your frames and, notably, can generate native audio to match.

"Flash" in the name means the fast tier, and the node leans into it - shot composition, prompt rewriting, and a soundtrack are all switches, not post-processing steps. It's the closest thing in this pack to "type, drop an image, get a finished clip with sound."

What you set

frameImages and positivePrompt are both required - this is I2V, and the first frame is the image you feed it.

• providerSettings.alibaba.audio - on by default: generates native audio aligned with the visuals. This is the killer feature and the biggest differentiator from the open Wan stack, which needs a separate audio pass. Off it costs less and runs faster.
• providerSettings.alibaba.promptExtend - LLM prompt rewriting, on by default. It improves results and breaks reproducibility, same trade as the other Alibaba nodes.
• providerSettings.alibaba.shotType - single or multi. Multi-shot tries to compose multiple shots into one clip; start on single.
• duration - 2 to 15 seconds (int, not a preset like some video nodes).
• size - a 13-option preset dropdown (720p/1080p across aspect ratios, plus "from input" modes that inherit the source's framing).
• audio - an optional STRING URL/UUID if you want to steer the soundtrack rather than let it generate.
• seed - exposed for reproducible takes.

Output is video (VIDEO socket) → video preview/save. Standard outputFormat/outputQuality/safety/ttl plumbing applies.

Installing

Part of the Runware/ComfyUI-Runware pack.

ComfyUI Manager: search Runware, install, restart.

Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/Runware/ComfyUI-Runware
pip install -r ComfyUI-Runware/requirements.txt

No weights to download (runware-sdk, pillow, soundfile). API key from runware.ai/api-keys (Settings panel or RUNWARE_API_KEY).

Where people get burned

Cost stacks up fast here: native audio, multi-shot, 15s duration, and numberResults > 1 all multiply the bill, and unlike a local Wan run there's no "free" step. Turn audio off and shotType to single for iteration, then add the extras once the visuals are locked. And remember the reproducibility caveat - promptExtend rewrites your prompt server-side, so a fixed seed doesn't guarantee a fixed result unless you disable it.