Runway Gen-4.5: Image to Video, Pricing and Gen-4 Comparison
Runway Gen-4.5 supports both text-to-video and image-to-video with 2–10 second control, stronger prompt adherence and current professional output options. Learn the real workflow, costs, limitations and how it compares with Gen-4 and competing tools.
Runway Gen-4.5 Video Generator
Runway Gen-4.5 at a glance
Generation modes
Text to video and image to video are both supported on the web and through Runway Dev.
Duration
Choose any whole-second duration from 2 to 10 seconds instead of only 5 or 10.
Cost
The published base rate is 12 Runway credits per output second.
Output
720p at 24 or 25 fps, with multiple aspect ratios for image-to-video.
Current model
What Is Runway Gen-4.5?
Runway announced Gen-4.5 in December 2025 as a newer general video model focused on motion quality, visual fidelity and precise prompt adherence. Unlike Gen-4 Video, which requires a first-frame image, Gen-4.5 supports both text-to-video and image-to-video. That makes it easier to begin with a written concept while preserving a controlled first-frame workflow when art direction already exists.
The model is designed to follow complex, sequenced instructions such as a subject action followed by a camera move and an atmospheric change. It supports durations from 2 through 10 seconds, allowing a small motion test without paying for a mandatory five-second result. Text-to-video uses 16:9 or 9:16; image-to-video supports those plus 1:1, 4:3, 3:4 and 21:9. Base output is 720p at 24 or 25 fps.
Gen-4.5 still has the familiar limitations of generative video. Cause and effect can appear in the wrong order, occluded objects may disappear, and requested failures may turn into successful actions. A strong model reduces these errors; it does not remove the need for shot design, review and post-production. Keep each generation centered on one coherent beat and inspect the entire clip, not only an attractive thumbnail.
| Specification | Gen-4.5 |
|---|---|
| Modes | Text to video; image to video |
| Duration | 2–10 seconds |
| Base cost | 12 credits per second |
| Output | 720p; 24 or 25 fps |
| Web access | Standard plan and above |
| API | Available as gen4.5 |
A 2-second test costs 24 credits; a 10-second generation costs 120 before optional professional-format surcharges.
Core workflow
Runway Gen-4.5 Image to Video: A Practical Workflow
In image-to-video mode, the source image defines the first frame: subject, composition, lighting, materials and style. The prompt should describe how that frame changes. Focus on subject motion, environmental motion, camera movement and timing. Avoid repeating visible details unless they must remain invariant. For the gallery image above, a useful prompt would describe how the fabric lifts and flows, whether the dancer moves, and whether the camera pushes in or stays locked.
Prepare the image at or near the target aspect ratio. Gen-4.5 image inputs must fall between 0.5 and 2.0 width-to-height, and the official output ratios include landscape, portrait, square and cinematic wide options. Cropping after upload can remove the negative space needed for movement, so compose the first frame with the camera path and subject destination in mind.
Use the shortest duration that can contain the action. Two or three seconds is enough to test a camera cue, fabric motion or expression change. Longer prompts that contain sequential actions need more time, but adding seconds does not fix an overloaded instruction. Break a complex scene into shots, generate each one, then edit them together. This also makes failures cheaper to isolate.
Iterate by changing one variable. If the camera is correct but the subject moves too fast, keep the image and camera phrase unchanged and adjust timing. If identity drifts, repair the source image or simplify the movement before adding more prompt detail. Save successful prompts with their exact source frames; a prompt without its image is not a reproducible image-to-video recipe.
- Describe motion rather than redescribing the image.
- Leave visual space in the direction the subject or camera will travel.
- Test the minimum duration before committing to a longer clip.
- Use one primary action and one clear camera instruction per shot.
- Review object permanence, hands, boundaries and the final frame before export.
- Keep the source frame and prompt together for reproducible iteration.
Upgrade decision
Runway Gen-4.5 vs Gen-4: What Is Actually Different?
The clearest functional upgrade is flexibility. Gen-4 Video needs an image and restricts duration to 5 or 10 seconds. Gen-4.5 can start from text or an image and accepts 2–10 seconds. Both have a 12-credit-per-second base rate, so a five-second clip costs the same 60 credits. Gen-4.5 provides more ways to begin and lets you pay for a shorter test.
Gen-4.5 is also Runway's current quality-focused model, with improvements in complex instruction following, motion, physical behavior and visual detail. That does not make every Gen-4 result obsolete. A simple first-frame shot may already work in Gen-4, while Turbo is much cheaper for broad iteration. Choose Gen-4.5 when text-only input, flexible timing or a difficult motion requirement has concrete value; use the older route when it already meets the brief or cost is the binding constraint.
| Feature | Gen-4 | Gen-4.5 |
|---|---|---|
| Text to video | No | Yes |
| Image to video | Yes; image required | Yes |
| Duration | 5 or 10 seconds | 2–10 seconds |
| Base rate | 12 credits/second | 12 credits/second |
| Frame rate | 24 fps | 24 or 25 fps |
| Position in lineup | Older controlled workflow | Current advanced general model |
| Best use | Designed first-frame shots | Flexible input and complex motion |
Workflow comparison
Runway Gen-4.5 vs CapCut, Kling AI and Seedance
These products are not interchangeable. CapCut is primarily an editing environment with integrated AI features; it is strongest when generation must flow directly into templates, captions, audio and social delivery. Kling AI is a dedicated generative platform known for several image/video control modes. Seedance is a model family available through different products and APIs, including Runway's own developer platform. Gen-4.5 is Runway's native current video model and integrates tightly with Runway sessions, assets and post-generation tools.
Compare the complete workflow, not a viral sample. Use the same source frame, equivalent duration and a prompt that tests your actual requirement. Score prompt adherence, identity and object stability, camera path, motion completion, export options, queue reliability and total cost after retries. Published pricing and free access change frequently, so the interface and checkout shown to your account should decide the purchase—not a static comparison table.
Choose CapCut when editing and distribution speed outweighs fine model control. Compare Kling when its specific controls or motion style fit the shot. Compare Seedance when API availability or its current video modes fit an existing pipeline. Choose Gen-4.5 when Runway's asset-to-generation-to-edit ecosystem, flexible 2–10 second duration and professional output options reduce handoffs.
| Option | Product shape | Strongest reason to test | Watch for |
|---|---|---|---|
| Runway Gen-4.5 | Native Runway video model | Prompt control and integrated workflow | 12 credits/second plus retries |
| CapCut | Editor with AI features | Fast edit, caption and delivery flow | Model availability varies by plan/region |
| Kling AI | Generative platform | Alternative motion and control modes | Compare equivalent settings and retry cost |
| Seedance | Model family across platforms | API and multimodal workflow options | Version and host affect specs and price |
The table compares workflow fit, not a permanent winner. Re-test with your own shot and current account terms.
Budget and access
Runway Gen-4.5 Pricing, Free Tier and API
Gen-4.5's base rate is 12 credits per generated second on both the web workflow and Runway Dev. A two-second test is 24 credits, five seconds is 60 and ten seconds is 120. The Standard plan currently includes 625 monthly web credits, but prices and plan inclusions can change. Count expected retries when budgeting; one accepted shot often requires more than one generation.
Runway's Free plan receives 125 one-time credits but only a limited, changing model set. Gen-4.5 officially requires Standard or higher in the web guide, so the free balance should not be presented as Gen-4.5 access. Free users also cannot buy add-on credits without upgrading. Check the model selector and upgrade message in the current account.
Runway Dev sells developer credits separately at $0.01 each. The gen4.5 API supports text-to-video and image-to-video for 2–10 seconds. Professional output options such as ProRes or PNG sequences add 5 credits per second, while supported 10-bit and HDR profiles have larger surcharges. Those formats are for production handoff, not a quality switch every user needs.
Aleph 2.0 solves a different job: it is a video-editing and transformation model rather than a synonym for Gen-4.5. A practical workflow may generate a shot with Gen-4.5 and then edit or transform it with Aleph 2.0. Compare them by task—new video generation versus editing existing footage—not by asking which model is universally better.
| Duration/format | Base Gen-4.5 cost |
|---|---|
| 2 seconds | 24 credits |
| 5 seconds | 60 credits |
| 10 seconds | 120 credits |
| ProRes or PNG sequence | Base rate + 5 credits/second |
| 10-bit/HDR profiles | Base rate + published format surcharge |
Web subscription credits and Runway Dev API credits are separate balances.
Direct answers
Runway Gen-4.5 Frequently Asked Questions
Continue comparing
Explore the complete Runway Gen-4 family
Return to the Gen-4 hub for References or use the Turbo guide for a focused speed-and-cost decision.