EvoLink MiniMax Hailuo 2.3 Fast (Official)

EvoLink_Hailuo23Fast is the budget door into MiniMax's Hailuo 2.3. Same strong motion expressiveness as the standard Hailuo node, but it's the "Fast" tier - cheaper and quicker - and it's image-to-video only. No text-to-video here; the first frame is required, and everything animates out from your image. If you're iterating on a single character or scene and want the Hailuo motion quality without paying full price per run, this is the one.

The trade is honest: it's the budget variant, so think drafts and rapid iteration, not your final polished render - though at up to 1080p it can pass for finished work at a glance.

The inputs that matter

• first_frame (required) - the IMAGE input. Connect Load Image here. This node has no text-to-video mode, so the frame is non-negotiable.
• prompt (required) - what happens and how the camera moves; Chinese-friendly.
• duration - 6 (default) or 10 seconds.
• quality - 768p (default) or 1080p.
• prompt_optimizer - defaults to on, and it rewrites your prompt before submission. If the output drifts from what you wrote, turn it off.
• fast_pretreatment - default off; shortens the optimization phase for speed.

Plus the pack-standard api_key (fill once, saved locally) and timeout_seconds (1200).

How it works

Identical plumbing to the rest of the pack: your frame uploads through EvoLink's file channel (auto-deleted after 72h, no third-party image host), the job posts to POST /v1/videos/generations, and the pack polls until the clip finishes, downloading it as a VIDEO object. Wire video into a native Save Video node. You also get task_id, status_info, result_urls (24h expiry), and response_json. Needs a recent ComfyUI for VIDEO output support.

Installing

No GPU, no model files, only requests:

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

Fully restart ComfyUI afterward, or use ComfyUI Manager and search EvoLink.

Troubleshooting

The pack-wide error set applies: 401 key, 402 balance, 403 access, 429 rate limit, and the VPN note - whitelist api.evolink.ai and files-api.evolink.ai for direct connection. The Hailuo-specific gotcha is prompt_optimizer rewriting your carefully-worded prompt; when the output does something you never wrote, that's your culprit. And remember this node is first-frame-only by design - if you find yourself wanting a last frame or text-to-video, you're on the wrong node; grab the standard Hailuo 2.3 node instead.