VIDEO
Runway Gen-3 on Elevence AI
Runway's Gen-3 Alpha video model — text-to-video and image-to-video with strong motion control.
Applies to web, iOS and Android
PROVIDER
Runway
MODALITY
video
PRICING
From $0.20 per second of generated video
Runway Gen-3 Alpha is one of the leading text-to-video and image-to-video models. It supports motion control, camera direction, and image-prompt conditioning — letting you generate short cinematic clips with directorial control over composition and motion.
On Elevence AI you can use Runway Gen-3 alongside Luma Dream Machine, Google Veo and Kling. Same credit pool, same API surface, easy switching between providers on the same scene description.
What Runway Gen-3 is great at
Strong motion realism and camera control
Runway Gen-3 produces convincing camera movement — pans, zooms, dolly shots — with motion that respects physical scene dynamics. Particularly strong for cinematic-style outputs.
Image-to-video conditioning
Drop in a still image as the first frame and Gen-3 animates from that anchor — useful for bringing existing brand assets, character art or photographs to motion without losing visual fidelity.
Director-friendly UX with motion brush
Runway's interface includes a motion-brush that lets you specify which parts of the scene should move and which should stay static. Combined with text prompts and reference images, you get directorial-level control.
Best for
•
Cinematic short clips and ad creative
•
Image-to-video animation of brand assets
•
Motion graphics and visual effects
•
Storyboard-to-video pipeline for pre-vis
Limitations
Short clip durations
Gen-3 outputs are typically 5–10 seconds. For longer narratives, generate multiple clips and stitch them in post (Elevence AI's multi-scene workflow helps here).
Cost compared to Luma
Per-second pricing on Runway is higher than Luma Dream Machine. For high-volume preview work, Luma may be more economical; reserve Runway for hero shots.
Code example
Drop-in OpenAI-compatible API. Swap the base URL and key, change the model field:
// Multi-scene video generation:
const session = await fetch('https://api.elevence.ai/api/v1/video/sessions', {
method: 'POST',
headers: { Authorization: `Bearer ${KEY}`, 'Content-Type': 'application/json' },
body: JSON.stringify({
model: 'runway-gen-3-alpha',
scenes: [{ prompt: 'Dolly shot through a misty pine forest at dawn' }],
duration_seconds: 6,
}),
});
Frequently asked questions
How long can a Runway Gen-3 clip be?
Single requests typically generate 5–10 seconds. For longer videos, Elevence AI supports multi-scene sessions: chain multiple clips with consistent characters/style and stitch the output.
Can I use an image as the starting frame?
Yes. Pass an image URL or base64 as the conditioning input — Gen-3 will animate from that anchor frame, preserving the visual style.
Runway vs Luma vs Veo on Elevence AI — which to pick?
Runway: best cinematic motion and camera control. Luma: most cost-effective per second, fast turnaround. Veo: best for realistic scenes with accurate physics. Try all three on the same prompt — Elevence makes side-by-side comparison easy.
What resolution does Gen-3 produce?
Up to 1080p depending on the variant. Resolution and duration affect cost; the live price is shown in the studio before generation.
Are generated clips usable commercially?
Per Runway's terms, yes — you own outputs and can use them commercially. Verify the latest Runway terms for any per-tier restrictions.
Related models
Try Runway Gen-3 on Elevence AI
Sign up free and access Runway Gen-3 alongside 60+ other models — one bill, pay-as-you-go.
Get started →