Runway
by Runway AI, Inc.
From the best-scoring video model to general world models that simulate and act in the world
Runway builds generative video and general world models, sold through a creative cloud app, a production API and a robotics toolkit. It is used by media and advertising teams who need cinematic video from text or images, and increasingly by gaming and robotics teams who need photorealistic simulation and synthetic training data.
Runway is the AI research company that spent five years as the best-known text-to-video tool and has, by 2026, repositioned around what it calls Real-World Intelligence: foundational models that understand, simulate and act in the world rather than merely render clips. Its February 2026 Series E raised $315 million led by General Atlantic at a $5.3 billion valuation, close to double the $3.3 billion it carried at its April 2025 Series D, with Nvidia, Adobe Ventures, AMD Ventures, Fidelity Management & Research, AllianceBernstein, Mirae Asset, Premji Invest and Felicis participating, explicitly to pre-train the next generation of world models across a team of roughly 140 people. The model line now spans Gen-4.5, released in December 2025 and the top-scoring video model in blind human preference testing at 1,247 Elo, a 21-point lead over Veo 3; GWM-1, its general world model; Solaris, an interface world model that generates interactive UI elements in real time; and GWM Worlds 2, which produces interactive 720p 24fps worlds with audio. Those models reach users through three distinct surfaces: Runway Creative, the all-in-one cloud platform for generating and editing video, images and audio that the company says serves more than 60 million creatives; Runway Dev, the API platform for production-scale deployment; and Runway Robotics, a toolkit combining policy inference, photorealistic simulation and synthetic data generation on GWM-1. Unusually, Runway also resells Google's Veo 3.1 inside its own platform, so customers get multi-model access without maintaining separate integrations.
The creative or production team at a media, advertising or gaming company that needs cinematic generated video with shot-to-shot consistency, plus an API mature enough to run it in production.
The highest-scoring video model in blind preference testing, reachable from both a finished editing app and a documented production API — and it fronts competitors' models too.
At a Glance
- Category
- Video & Animation
- Pricing
- Freemium, Subscription, Usage-based, Contact for pricing
- Target Market
- Creative Directors, Enterprise Developers, Marketing Leaders, Game Developers, Robotics Engineers
- Deployment
- Cloud-only, API-based
- Team Size
- 101-250
- Customers
- 60M+ creatives on Runway Creative (company claim)
Key Features
- ✓Gen-4.5 video model
Released December 2025 and the highest-rated video model in blind human preference testing at 1,247 Elo, with state-of-the-art motion quality and prompt adherence.
- ✓GWM-1 general world model
A foundational model for simulating and acting in physical environments, powering the robotics toolkit's policy inference and synthetic data generation.
- ✓Solaris interface world model
Generates interactive user-interface elements in real time, extending world modelling from physical scenes to software surfaces.
- ✓Runway Dev API
Production API that independent reviewers rate the most thorough in the category for documentation, covering webhooks, batch processing and error handling.
- ✓Multi-model access including Veo 3.1
Competitor models are available inside the Runway platform, so a customer can switch models without maintaining a second vendor integration.
- ✓Studio delivery pipeline
Unlimited 4K upscaling on Pro, plus HDR and ProRes studio-grade outputs, custom voice clones and Brand Kits on higher tiers.
Capabilities
Use Cases
- •Advertising and campaign video
Generate and iterate on cinematic ad concepts in hours rather than commissioning a shoot, then deliver in ProRes for finishing.
- •Previsualisation and storyboarding
Turn scripts and reference stills into moving previs so directors and clients approve a shot before any crew is booked.
- •Synthetic data for robotics
Use GWM-1 photorealistic simulation to generate training data and run policy inference without building physical test environments.
- •In-product video generation
Embed Runway Dev in a consumer or SaaS product so end users generate video, with batch processing and webhooks handling production load.
- •Game world prototyping
Generate interactive 720p worlds with audio from GWM Worlds 2 to test environment concepts before committing art and engineering effort.
Ideal For
Best For
- ✓Advertising, film and marketing teams generating cinematic video from text or reference images with consistent characters across shots
- ✓Product teams embedding video generation in their own application via a documented API with webhooks, batch processing and error-handling patterns
- ✓Studios that want multi-model access — including Google's Veo 3.1 — through a single vendor integration rather than several
- ✓Robotics and gaming teams needing photorealistic simulation and synthetic training data from a general world model rather than hand-built environments
- ✓Post-production workflows requiring 4K upscaling, HDR and ProRes studio-grade delivery outputs
Not Ideal For
- ✗Teams that need tightly forecastable unit costs — Runway meters in credits rather than per second, and independent API comparisons call that materially harder to budget than Sora 2's or Veo 3.1's per-second billing
- ✗High-volume pipelines whose prompts brush content filters, because moderated and blocked generations still consume credits
- ✗Buyers who need audio generated natively alongside video; Runway's audio path runs through a separate Lip Sync feature gated to Pro plans, where Veo 3.1 synthesises audio by default
- ✗Cost-sensitive creators generating at volume, where Kling at roughly $10-40 a month undercuts Runway's $95 Max tier for broadly comparable output
Integrations
Deployment
Market & Ratings
60M+ creatives on Runway Creative (company claim)
Market Analysis
Pros
- ✓Measurably the strongest video model on blind human preference testing, not merely on vendor benchmarks
- ✓Gen-4.5 produces more consistent, predictable results across generations than rivals that win on single-shot spectacle but vary more run to run
- ✓Unusually well-documented API — webhooks, batching and error handling — which is what separates a demo from a production pipeline
- ✓Multi-model access inside one platform reduces integration cost when the frontier moves, as it repeatedly has
- ✓Heavily capitalised at a $5.3B valuation with Nvidia, AMD and Adobe on the cap table, so the compute and distribution story is credible
Cons
- ✗Credit metering is harder to forecast than per-second billing, and blocked or moderated generations still consume credits — a real cost for filter-adjacent prompts
- ✗No native audio bundled with video; Lip Sync is a separate feature gated to Pro, where Veo 3.1 synthesises audio by default
- ✗Expensive against the low-cost field — Hacker News practitioners contrast Runway's $95-a-month top tier with Kling at $10-40
- ✗Competitive pressure is intense and two-sided: Sora 2 and Veo 3.1 on video, and World Labs plus Google DeepMind now shipping public world models
- ✗No retrievable G2, Capterra or TrustRadius rating, and Hacker News interest has thinned markedly since the Gen-4 launch at 79 points to single digits for Gen-4.5
- ✗No published security certifications, on-premises option or data-residency commitment, which limits it in regulated environments
Pricing
Free
$0
- ✓125 one-time credits
- ✓5GB storage
- ✓Limited model selection
- ✓Watermarked output
Standard
From $12/mo
- ✓625 credits per month
- ✓All video and image models
- ✓5 parallel generations
- ✓20GB storage
- ✓No watermarks
- ✓$15 billed monthly
Pro
From $28/mo
- ✓2,250 credits per month
- ✓15 parallel generations
- ✓Unlimited 4K upscaling
- ✓1 custom voice clone and 1 Brand Kit
- ✓100GB storage
- ✓$35 billed monthly
Max
From $76/mo
- ✓9,500 credits per month
- ✓20 parallel generations
- ✓Early access to new models
- ✓HDR and ProRes studio-grade output
- ✓3 voice clones and 3 Brand Kits
- ✓500GB storage
- ✓$95 billed monthly
Enterprise
Contact for pricing
- ✓Custom volume licensing
- ✓Enhanced security and support
- ✓Custom implementations
Everything is metered in credits rather than seconds of output, which is the single most important commercial fact about Runway: independent API comparisons put the effective rate at roughly $0.05-0.15 per second depending on model and resolution, versus Sora 2's flat $0.10 per second and Veo 3.1's $0.15-0.60, and note both that credits are harder to forecast and that moderated or blocked generations still burn them. List tiers run Free with 125 one-time credits, Standard at $12 a month, Pro at $28 and Max at $76 when billed annually, or $15, $35 and $95 monthly, a flat 20% annual discount; Enterprise is quote-only. At production scale the same comparison models 500 eight-second 1080p videos a day at $12,000-18,000 a month for Runway, against roughly $48,000 for Veo 3.1 Standard and about $18,000 for Veo 3.1 Fast. Audio is not bundled: Lip Sync sits behind the Pro tier, and voice cloning is capped at one clone on Pro and three on Max.
Security & Compliance
Sources
This page was written from 5 sources, 3 on domains other than runway.com.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
Subscribe