Video Models

Here are the video models currently available in FLORA.

COMING SOON: Model reference cards for each of our models.

Model	Est. time (s)	Credits	Description	Best For
Aleph (Gen-4)	220	1000	Runway’s flagship Gen-4 engine; delivers high-consistency, cinema-grade frames.	Storyboards, pitch films, and director “look reels” where every frame must hold up as a polished production still and camera continuity is essential.
Act-Two (Gen-3)	180	750	Character-centric Gen-3 model with strong identity tracking.	Turning live-action reference into stylised animation or explainer shorts—keeps faces on-model and emotions readable across multiple shots.
Animatediff	300	100	Real-time latent-diffusion animator.	Rapid motion-graphics ideation, looping GIFs, and meme-style clips where cost and turnaround trump photoreal fidelity.
Hailuo Minimax	150	760	Multimodal model with nuanced prompt comprehension.	Narrative shorts (30 – 60 s) that need coherent character arcs, evolving lighting cues, and tight match between text instructions and on-screen action.
Minimax 02	90	250	Turbo prototype variant of Hailuo.	Ultrafast generation of vertical reels, bumper ads, or A/B variants when you need feedback-ready footage in minutes.
Veo 2	240	900	Google’s first production Veo release; cinematic motion, occasional drift.	Hero shots and aerial moves—trailers, architecture fly-throughs, luxury product reveals—when you can polish minor scene inconsistencies in post.
Veo 3	260	950	Latest Veo with stronger scene coherence.	4 K trailer sequences, high-end commercials, or narrative teasers where lighting, subject placement, and camera blocking must survive detailed review.
WAN 2.2	210	480	Open-source T2V/I2V/IV2V model from Alibaba.	Turning still images into subtle parallax or 360° spin videos—great for catalog, e-commerce, and editorial assets needing controlled motion paths.
Kling Pro 1.5	300	667	High-quality, stylistically diverse Pro engine.	Music-video prototypes, mood pieces, and concept reels that rely on dramatic grading and ambitious camera choreography.
Kling Pro 1.6	300	367	Photoreal upgrade with advanced lighting & physics.	Product demos, architectural walk-throughs, and people-in-scene renders that must feel believably lit and physically grounded.
Kling 2.0	300	667	Master-grade T2V accepting mixed inputs & 60+ styles.	Complex long-form sequences combining text, reference images, and scratch audio—ideal for cinematic pre-viz or style explorations.
Kling 2.1	330	600	Pro variant focused on first/last-frame fidelity.	TikTok-style transitions, before-and-after promos, or any edit where the final frame must match a supplied still exactly.
Lightricks LTXV	90	100	Fast generator tuned for flicker-free transitions.	Storyboards, animatics, and motion-study clips where fluid scene changes are more important than photoreal detail.
Luma Ray 2	200	947	Large-scale model for natural, coherent motion.	Lifestyle, nature, or travel footage—drone-style glides, fashion runways, or B-roll sequences that demand smooth, realistic camera flow.
Luma Ray 2 Flash	50	320	Speed-optimised Ray 2.	Quick variant testing, social-media cut-downs, and preview edits when you want Ray-quality motion without Ray-level cost or wait.
Luma Modify Video	200	2334	Restyles existing footage while preserving motion.	Day-for-night conversions, illustrative repainting, or “live-action → anime-look” transformations that keep timing intact.
Marey Motion Transfer	400	2668	Transfers motion from a reference clip.	Applying a dancer’s choreography, an athlete’s swing, or a steadicam path to new CG characters or scenes for pre-viz or VFX.
Marey Pose Transfer	300	2668	Frame-accurate pose transfer tool.	Digital-double pose matching, stunt-viz, or freeze-frame-to-motion ads that hinge on exact body-pose replication.
Marey (i2v)	420	2000	High-fidelity image-to-video generator trained on licensed data.	Brand-safe commercial spots and legal-clear hero sequences built entirely from approved still assets.
Pika	90	250	Character-focused T2V with expressive faces.	Avatar-led explainers, talking-head shorts, and emotional micro-stories that depend on consistent facial performance.
Seedance 1.0 Pro	120	827	Versatile CG-style Pro model with strong text adherence.	Premium social ads, stylised title sequences, and cinematic game-trailer mood boards where painterly lighting is a plus.
Tencent Hunyuan	180	680	13 B-param T2V with video-to-audio pairing.	Multi-locale ad campaigns and culturally nuanced content requiring language-specific versions and synced soundtrack stubs.