MiniMax's fast video model, with sound on every clip
H3 Max is MiniMax's H3 video model, post-trained by fal for fast generation. It renders 5 to 15 seconds at 480p or 768p, generates a soundtrack with every clip, and expands your prompt before rendering — a balanced pass by default, a more careful one in quality mode, or none at all when you want your wording followed exactly.
Plan limits apply on top of these — see pricing.
Written for this model in particular. Open one in the composer and edit it from there.
“A street-food cook tossing noodles over a roaring wok, sparks and steam, handheld close-up”
Try this prompt“Waves breaking over black volcanic rocks at dawn, slow push-in toward the spray”
Try this prompt“A paper lantern drifting up over a crowded night market, the camera tilting up to follow it”
Try this promptLong-form AI video with reference control and native audio
ByteDance4K AI video with a lockable camera and multi-subject references
ByteDanceThe cheapest way to test a video idea before committing credits
GoogleGoogle's flagship video model, with synchronized audio at up to 4K
GoogleVeo 3.1 quality at lower latency, for iterating on a shot
Black Forest LabsBlack Forest Labs' first video model, with a fast draft mode
20 free credits on sign-up, no credit card required.