Runway Upscale V1: the still is the contract (5cr)

Runway Upscale V1 is image-to-video. If the label is wrong, every second is waste.

Read that claim against the catalog, not against a generate button. Runway Upscale V1 is content type video_upscale, category video_upscale, requires_image true, 5 credits, audio empty, no duration list, no max output resolution. The description: "AI image upscaling model by Runway for enhancing image resolution while preserving detail." Features: image_upscaling, resolution_boost, detail_preservation. If you came here to invent a scene, the label is already wrong, and every second of source you feed it is waste.

video_upscale is the catalog word

This is not text-to-video. This is not a 5s/10s clip generator. Runway's generate rows live on the Runway roster next to this one. Gen-4.5 and friends invent pictures. Upscale V1 takes a file you already have and tries to preserve detail while boosting resolution.

requires_image: true is the gate. No plate (or no source file), no job. The "still is the contract" line is literal: upscaling cannot recover a label that was never in the pixels. It can sharpen what is there. It can also sharpen compression.

The task surface is upscale a video to 4K or upscale an image, depending on the file. The glossary term is upscaling. Best Runway model is the family ranking — useful if you have not yet admitted you need post, not a generate.

Five credits on a file you already have

5 credits is the catalog credits field. The price matrix bills per second of duration in credits, with a published min/max range; the live quote is the number that charges. None of those figures are USD. All of them assume an input.

There is no 5s/10s picker because this row is not choosing a clip length to invent. Length is the length of the source. If the source is the wrong scene, you are paying to make the wrong scene larger.

Do not prompt a new product into an upscaler. There is no prompt that replaces the file. Detail preservation is the feature. Invention is the anti-feature.

Sharper is not newer

People land on Upscale V1 because the brand is Runway and the output looks "higher quality" in a screenshot. Higher quality of what? If the what does not exist yet, you wanted a generate: AI video generator or image-to-video. If the what exists and is soft, you wanted this.

The input is a picture-like file and the output is that file at higher resolution. That does not make it a scene generator. It makes it post. Treat a missing source like a missing still: stop, go get the file, then spend the 5 credits.

Runway is the only brand on the provider hub whose models include a dedicated video upscaler. That uniqueness is the reason the row exists. It is also the reason not to confuse it with Gen-4.

Preserve, do not invent

If the still (or clip) is signed and too small, upscale. If the still is unsigned, generate. If the still is signed and the motion is wrong, that is a different model entirely. Upscale V1 will not restyle, will not lipsync, and will not write a soundtrack. Audio is empty. Resolution is unspecified as a ceiling — judge the result at full size, because artifacts get bigger too.

FAQ

Is Runway Upscale V1 a text-to-video model?

No. Content type and category are video_upscale. Description is image upscaling with detail preservation. 5 credits, requires an input, no duration list.

Can I use it to create a clip from a prompt?

No. requires_image is true. A prompt without a file is the wrong label. Every second of a generate you do not have is not something this row can spend.

Why are there no 5s / 10s options?

Because this row does not generate a clip length. It enhances a source. Length comes in with the file.

Will it make a soft logo readable?

It enhances existing detail. It does not typeset a logo that was mush in the source. Fix the still on an image model, then upscale if you still need the resolution boost.