Download the Pict.AI iOS App, Free

Runway AI: how Gen-4 works and what video generation costs

Runway AI is a generative-media platform for creating videos and images, with editing and compositing tools in the same workspace. Runway Gen-4 creates video from an input image and a text prompt describing motion. It costs 12 credits per generated second: 60 credits for five seconds or 120 for ten. For practical results, start with a strong reference image, generate short variations, and inspect the motion before committing more credits.

At a glance
What it isA generative-media platform, also known as Runway ML
Gen-4 video inputAn input image plus a text prompt
Gen-4 video cost12 credits per generated second
Example generation costs5 seconds: 60 credits; 10 seconds: 120 credits
Gen-4 Image cost720p: 5 credits; 1080p: 8 credits
Useful starting workflowPrepare one reference image and generate short motion variations

What Runway AI does - and which model to choose

Runway AI combines generative video, image generation, editing, and compositing tools. Runway ML is another name associated with the platform; it does not describe a separate video model. Within the workspace, the model matters more than the platform name because it determines your inputs, generation settings, and credit consumption.

Runway Gen-4 uses an input image and a text prompt to generate video. The image establishes the starting composition, subject, lighting, and visual style. The prompt then directs what moves and how the camera behaves. This makes Gen-4 a practical option when you already have a product photograph, character illustration, or carefully composed scene that you want to animate.

Gen-4.5 is also part of Runway's video model family. Its listed generation rate is 12 credits per second. Do not assume that identical credit rates mean identical inputs, output settings, or access: select the model in the workspace and inspect its available controls before planning a batch of shots.

Separate image generation from video generation in your budget. Gen-4 Image costs 5 credits for a 720p image or 8 credits for a 1080p image. Those are image-generation settings, not a statement of Gen-4 video's output resolution. If the starting image is already wrong, fix it before paying to animate it.

Runway Gen-4 compared with Gen-4.5 and PixVerse V6

A useful comparison separates model cost from workflow. Runway's video rates below are measured per generated second; its image rates are measured per generated image. PixVerse uses its own credits, so a lower numerical credit rate does not establish a lower dollar cost.

Model or toolRelevant functionCredit costDecision point
Runway Gen-4Video from an input image and text prompt12 credits/secondUse when the starting composition is already defined
Runway Gen-4.5Video generation12 credits/secondCheck the model's available inputs and settings before choosing
Runway Gen-4 ImageStill-image generation5 credits at 720p; 8 at 1080pBudget reference-image creation separately
PixVerse V6Text-to-video and image-to-video; 1-15-second clipsAt 1080p: 18 credits/second without audio; 23 with audioConsider when selectable generated audio is part of the brief

PixVerse V6 supports 360p, 540p, 720p, and 1080p output. Its audio option changes the generation cost, so compare the exact settings rather than just the model names. A ten-second V6 clip at 1080p uses 180 credits without audio or 230 with audio.

If you are also considering vidu ai, compare the same reference image, action, duration, and intended delivery format rather than unrelated showcase clips. Across services, the useful measure is the cost and effort required to produce an acceptable shot, including rejected attempts, not simply the price of one generation.

How to animate an image with Runway Gen-4

Start with one shot, not a complete sequence. A five-second clip with one clear action is easier to assess than a longer generation containing several movements, camera changes, and story beats.

  1. Prepare the reference image. Use a clear subject, intentional framing, and enough space for the planned movement. Resolve visible defects in faces, hands, packaging, or background objects first.
  2. Open the video workflow and select Gen-4. Upload your reference image. Confirm the selected model before generating, since settings and costs belong to the model rather than to the workspace as a whole.
  3. Describe the motion. State the subject action and camera behavior in plain language. For example: “The subject slowly turns toward the window. The camera remains stationary. The curtain moves gently.”
  4. Choose a short duration. A five-second Gen-4 generation costs 60 credits. Use that length to check whether the central movement works before committing to a ten-second attempt.
  5. Generate and change one variable. If the result is wrong, revise the action or camera instruction rather than rewriting everything at once. Keep the reference image unchanged while diagnosing motion.
  6. Inspect, edit, and export. Review the beginning, middle, and end of the clip. Use editing or compositing tools where available, then check the exported file for artifacts and suitability for publication.

For reference preparation, Pict.AI is one option: an AI photo editor app for iPhone and Android, plus a website with guides and free image tools. It fits the still-image stage; Runway handles the video-generation stage in this workflow.

Runway credit costs and a realistic generation budget

As of September 2026, Gen-4 video generation costs 12 credits per second. The arithmetic is straightforward, but iteration changes the budget quickly. Each new generated attempt adds to the amount you need to allocate.

Generation taskCalculationCredits
One five-second Gen-4 clip5 × 1260
One ten-second Gen-4 clip10 × 12120
Four five-second attempts4 × 60240
Six ten-second attempts6 × 120720
Four 1080p Gen-4 images plus four five-second videos(4 × 8) + (4 × 60)272

For a three-shot edit, allowing four five-second attempts per shot means budgeting 720 credits for video generation. The selected footage totals 15 seconds, but the generated footage totals 60 seconds. That distinction is central to estimating a project: you pay for generation, not only for the footage that survives the edit.

Runway offers credit-based subscription plans. Before subscribing, compare the dollar price, included credit balance, model access, and any separate charges for the tools you intend to use on the Runway pricing page. A credit rate alone is not a dollar-per-video quote.

Set an attempt limit before starting. For example, allow three short motion revisions, then return to the reference image if the same defect persists. Repeatedly generating an unsuitable composition can consume more credits without addressing the actual problem.

Video defects to check before accepting a generation

A plausible opening frame does not guarantee a usable clip. Treat these as inspection targets for generated video, not as quantified failure rates for a particular Runway model.

  • Identity and appearance changes. Check whether the face, hairstyle, clothing, and accessories remain consistent as the subject turns or moves. Compare the final frame with the reference image.
  • Hands and object contact. Inspect fingers, handles, tools, and the point where a hand meets an object. A convincing pose can still contain a broken grip or an object that changes shape.
  • Product labels and text. Read logos and packaging throughout the clip, not only at the start. For precise commercial lettering, consider adding the final text during editing instead of relying on generated movement to preserve it.
  • Physics and spatial continuity. Watch feet, shadows, reflections, and objects passing behind other objects. Look for sliding, unexplained acceleration, or elements that disappear and return.
  • Camera behavior. Check whether the shot follows your requested stationary frame, pan, or push-in. Unwanted camera motion can make an otherwise acceptable subject animation unusable.

When a defect appears, identify whether it comes from the starting image, the requested action, or the amount of movement. A simpler action or a better-framed reference may be more useful than another unchanged attempt.

Visual quality is only one publication check. Confirm permission to use source images, identifiable people, and protected assets. Review the applicable commercial-use terms, and avoid presenting a generated or materially altered event as authentic footage.

Prompt and editing tips for more usable Runway clips

Use the reference image to carry appearance and the prompt to carry movement. Repeating every visible detail can obscure the instruction that matters. A concise prompt such as “The camera slowly pushes toward the vase while the flowers sway gently” gives the shot a clearer purpose than a paragraph combining lighting, mood, multiple actions, and several camera moves.

Assign one primary action to each shot. If a character must pick up a cup, turn, walk away, and speak, break the sequence into separate shots and assemble them afterward. This also makes replacements cheaper: you can regenerate one weak moment instead of replacing the whole sequence.

Match the movement to the framing. A tightly cropped portrait leaves little room for walking or broad gestures. A product close-up may benefit from restrained movement rather than a large orbit that demands views absent from the starting image. Compose the reference for what the shot needs to reveal.

Keep a small generation log with the reference filename, prompt, model, duration, credit cost, and rejection reason. It helps distinguish a successful revision from a merely different result. After selecting footage, trim weak beginnings or endings and add titles, sound, and transitions as separate editorial decisions.

Finally, verify the delivery settings for the selected model rather than assuming all Runway tools export identically. Consult the Gen-4 creation guide for its controls. Judge the finished export at its intended viewing size, including vertical mobile playback when relevant.

Runway AI: how Gen-4 works and what video generation costs

Frequently asked questions

What is Runway AI used for?

Runway AI is a generative-media platform used for video generation, image generation, editing, and compositing. Its Gen-4 video model animates an input image using a text prompt that describes movement. A practical workflow is to prepare a reference image, generate several short variations, select usable footage, and finish the sequence with editing.

Are Runway AI and Runway ML the same thing?

Runway AI and Runway ML refer to the same platform, rather than two competing services. Gen-4 and Gen-4.5 are model names within that platform. When following a tutorial or estimating costs, check the specific model being used: the platform name alone does not tell you which inputs, settings, or credit charges apply.

Does Runway Gen-4 need an input image?

Yes. Runway Gen-4 video uses an input image together with a text prompt. The image establishes the starting scene, while the prompt directs subject movement and camera behavior. Prepare the image before generating: visible defects, unsuitable framing, or insufficient space for movement can undermine the shot even when the motion instruction is clear.

How many credits does Runway Gen-4 cost?

Gen-4 video costs 12 credits per generated second. A five-second generation therefore costs 60 credits, and a ten-second generation costs 120. Four five-second attempts total 240 credits. Gen-4 Image has separate charges: 5 credits for a 720p image or 8 credits for a 1080p image.

How much does Runway AI cost per month?

Runway uses credit-based subscription plans, so the monthly bill depends on the plan you select. Compare its dollar price, included credits, and access to your intended model before subscribing. For estimating usage, Gen-4 costs 60 credits per five-second generation; ten such attempts require 600 credits, before image generation or other separately charged work.

What is the difference between Runway Gen-4 and Gen-4.5?

Gen-4 and Gen-4.5 are separate video models in Runway's model family. Both have a listed generation rate of 12 credits per second. Gen-4 uses an input image and text prompt. Equal pricing does not establish equal capabilities, so inspect Gen-4.5's available inputs, duration settings, output options, and plan access before selecting it.

Why does my Runway video change faces or distort objects?

Generated video can lose visual consistency as movement introduces new poses, angles, or object interactions. Inspect faces, hands, labels, and contact points throughout the clip. Try a clearer reference image, less movement, or one action per shot. If the same defect repeats, change the composition or action rather than spending credits on identical attempts.

Can I use Runway AI videos commercially?

Before commercial publication, check the terms that apply to your Runway account and chosen tools. Also confirm rights to uploaded photographs, artwork, logos, and identifiable likenesses. Permission to generate or download a clip does not by itself resolve third-party rights. Review the output for misleading representations and inaccurate product details before using it in advertising.