The short answer (August 2026): Runway is a web-based AI video platform where you supply a text prompt or a still image and get back a short cinematic clip, with no local software required. Its current flagship model, Gen-4.5, generates between 2 and 10 seconds of video at 12 credits per second; paid plans start at $15 per month (or $12 per month billed annually) for 625 credits, while the free plan gives you a one-time 125-credit allotment with watermarked output. Runway now hosts third-party models including Google Veo 3.1 and ByteDance Seedance 2.5 inside the same credit system, so it functions more like a multi-model video marketplace than a single tool. Sign up at runway.com, select a model, type a prompt, and download your clip.
Last verified: August 25, 2026, against official docs and pricing pages.
What Runway is now
Runway launched as a video editing and effects suite for independent filmmakers. Since 2023 it has pivoted its core product to AI video generation: you describe a shot in text or attach a reference image, and a generative model synthesizes a clip that matches your description.
The platform now runs on three layers. The first is Runway's own model family: Gen-4.5 for cinematic generation, Gen-4 Turbo for fast budget drafts, Act-Two for character animation, and Aleph 2.0 for editing existing footage. The second layer is third-party models licensed into the same interface, including Google Veo 3.1 and ByteDance Seedance 2.5, each pulling credits from your Runway balance. The third layer is production tooling: a Studio editor for assembling and exporting sequences, a Runway Agent for multi-step marketing campaign creation from conversational prompts, and a Runway MCP server that connects the platform to external AI agents.
If you are starting out, none of the third-party models or agentic features matter yet. The default workflow is: open Gen-4.5, type a prompt, generate a 5-10 second clip.
Plans, credits, and real cost per video
Runway bills by credits. Every second of generated video pulls from your balance, with the per-second rate depending on the model. Credits do not carry over between months on Free, Standard, or Pro plans.
| Plan | Monthly price | Annual price | Credits/month | Notable features |
|---|---|---|---|---|
| Free | $0 | $0 | 125 (one-time, never renews) | Selected models only; all output watermarked |
| Standard | $15 | $12 | 625 | All models; no watermarks; 4K upscaling |
| Pro | $35 | $28 | 2,250 | Custom voices for lip sync and TTS; 500GB storage |
| Max | $95 | $76 | 9,500 | One-month credit rollover; ProRes and image sequence export; 16-bit HDR; first access to new models |
| Enterprise | Custom | Custom | Custom | SSO, workspace analytics, configurable teamspaces, priority support |
What those credits buy in practice: the Standard plan's 625 credits cover about 52 seconds of Gen-4.5 footage per month, roughly 10 five-second clips. Switch to Gen-4 Turbo (5 credits/sec) and the same 625 credits produce about 125 seconds total. Pro's 2,250 credits stretch to about three minutes of Gen-4.5 or seven and a half minutes of Gen-4 Turbo. If you plan to generate more than a handful of clips per week, Standard will constrain you fast.
The API charges $0.01 per credit directly, so a five-second Gen-4.5 clip costs $0.60 per generation. Full pricing details, including the API rate card, are at docs.dev.runwayml.com/guides/pricing and the subscription plan comparison is at runway.com/pricing.
Your first video: step by step
These steps apply to Gen-4.5, the current default model on the platform.
1. Create an account. Go to runway.com and sign up with email or Google. Free plan credits appear in your balance immediately on signup.
2. Open the video generator. From the dashboard, select Gen-4.5 under Video Generation. You will see fields for mode (text-to-video or image-to-video), prompt, duration, and aspect ratio.
3. Choose your input mode. Text-to-video generates a clip from a description alone. Image-to-video takes a still and animates it according to your prompt. Image-to-video tends to produce more controlled framing when you have a reference composition worth preserving.
4. Write your prompt. For text-to-video, describe the subject, setting, and motion in the same prompt. For image-to-video, describe only what should move; re-describing visual content already in the image tends to suppress motion rather than clarify it.
5. Set duration and aspect ratio. Gen-4.5 generates between 2 and 10 seconds per clip. Aspect ratio options include 16:9 (landscape), 9:16 (vertical, for Reels or Shorts), 1:1, 4:3, 3:4, and 21:9 widescreen. Set the ratio before generating; changing it after requires re-cropping the source image.
6. Generate and review. A five-second Gen-4.5 clip costs 60 credits and typically completes within a couple of minutes. Runway accepts follow-up feedback prompts to refine a result without spending full credits on a new generation from scratch.
7. Download or continue in Studio. Clips export directly from the generation view. Open them in the Studio editor to trim, combine multiple clips, add audio from Runway's audio tools, and export a finished sequence.
The model landscape on Runway
Gen-4.5 is the right starting point for most work. The other models on the platform serve specific tasks at meaningfully different credit costs, and switching between them costs nothing beyond the credits consumed.
| Model | Best for | Credits/sec | What to know |
|---|---|---|---|
| Gen-4.5 | Cinematic clips, complex prompts, character consistency | 12 | Supports single-reference-image character consistency across shots |
| Gen-4 Video Turbo | Draft iterations, high-volume generation | 5 | Lower visual quality; useful for testing prompts before committing full credits to a Gen-4.5 generation |
| Act-Two | Character animation from a performance video | 5 | Requires Standard plan or higher; transfers head movements, facial expressions, and upper-body gestures from a driving video onto your character image |
| Aleph 2.0 | Frame-level editing of existing footage | 28 (56-credit minimum) | Changes applied to a single frame propagate automatically through the full clip |
| Google Veo 3.1 | Native audio-embedded video generation | 20 (no audio) / 40 (with audio) | Audio generation doubles the per-second credit cost |
A practical note on audio: Gen-4.5 generates video without embedded audio. Runway's Seed Audio tool generates speech, sound effects, or music from a text description and can be layered onto any clip after generation.
Prompting for results that hold up
The most important rule: image-to-video prompts describe motion; text-to-video prompts describe the scene and the motion together.
For image-to-video, focus entirely on what should change or move. Describing visual elements already present in the image gives the model redundant information and tends to reduce movement. Write "she slowly turns her head to the left, hair drifting slightly" instead of "a woman in a red jacket slowly turns her head to the left." The jacket is visible; the model does not need to be told.
For text-to-video, include at minimum a subject, a setting, a lighting quality, and one camera instruction. Standard cinematography terms work reliably with Gen-4 models: pan, tilt, dolly in, dolly out, truck left, truck right, orbit, crane, handheld, steadicam, locked-off. A few rules that hold up consistently:
One camera action per clip. Compound instructions such as "zoom in while panning left and tilting down" produce confused motion. Pick the single most important move and describe it in isolation.
Add a speed modifier. "Slow dolly in" behaves differently from "dolly in" with no qualifier. Without a modifier the model picks an arbitrary default.
Describe what the camera reveals during dramatic movement. "The camera cranes up slowly to reveal the full skyline beyond the rooftop" gives the model a clear sense of spatial progression through the shot, which produces better motion than "crane shot of a skyline."
Three tools beyond basic generation
Act-Two transfers a performance from a driving video onto a character image or existing clip. You upload a short video of someone acting out a scene, add a character reference image, and Act-Two applies the full motion capture: head position, facial expressions, and hand and upper-body gestures. The result is an animated clip at 5 credits per second, making it the most credit-efficient way to get expressive character movement on the platform. Act-Two requires Standard plan or higher. Full documentation, including input video requirements, is in the Performance Capture article in Runway's help center.
Aleph 2.0 edits existing footage at the frame level. You select specific elements in a frame to change and the model propagates those changes consistently across the full clip without manual frame-by-frame work. At 28 credits per second with a 56-credit minimum, it is the most expensive model on the platform per second. Use it on footage where re-shooting the original would cost more than the credit spend.
Studio, Runway's native editing interface, handles trimming, reordering, and combining clips, adding audio tracks, and exporting finished sequences. It is available on all paid plans and removes the need for a round-trip to an external editor for basic assembly. For projects that need only cut-and-join editing, it covers the full workflow inside Runway.
If you want to connect Runway to an external AI workflow, the Runway MCP server lets agents such as Claude or ChatGPT call Runway models directly. Setup is similar to other MCP integrations; see the AI Weekly MCP server guide for the general pattern.
FAQ
Does Runway offer a free trial?
The free plan gives you 125 credits on signup that do not expire and do not renew. At 12 credits per second, that covers about 10 seconds of Gen-4.5 video total across your entire free account. All free-plan output includes a visible Runway watermark. Paid plans starting at $15/month remove the watermark and unlock all models.
How many videos can I generate per month on the Standard plan?
Standard gives you 625 credits per month. At 12 credits per second on Gen-4.5, that is roughly 52 seconds of total video, about 10 clips at five seconds each. Switching to Gen-4 Turbo at 5 credits per second extends the same budget to about 125 seconds, or 25 five-second clips. Credits expire at the end of each billing cycle on Standard.
Can I access Runway via API for my own application?
Yes. The Runway developer API at docs.dev.runwayml.com supports Gen-4.5, Gen-4 Turbo, and other models at $0.01 per credit. A five-second Gen-4.5 clip costs $0.60. API billing is separate from subscription credits and requires its own payment setup.
Do unused credits roll over if I do not use them all in a month?
On Free, Standard, and Pro plans, unused credits expire at the end of each billing cycle. The Max plan ($95/month, or $76/month billed annually) includes a one-month rollover, so credits left unused in a given month carry into the following month only.
What is the longest video I can generate in a single clip?
Each Gen-4.5 generation produces up to 10 seconds of video. For longer sequences, generate multiple clips and assemble them in the Studio editor. Runway Agent can orchestrate multi-shot commercial content from a single conversational prompt, though each underlying generated clip remains capped at 10 seconds per generation.
If this kind of practical breakdown saves you time, get AI Weekly free, 3 issues a week, read by 40,000+ practitioners.