Runway Gen-4.5 is silent 50 credits

Runway Gen-4.5 is 50 credits, silent, 5 seconds or 10 seconds. You are buying control — cinematic motion, camera, a short generate you intend to specify — not native audio. Audio is false on the published card. Categories are text-to-video and image-to-video. An image is not required. Resolution is unset on the catalog row; do not invent 1080p or 4K. Fifty credits is the sticker. It is not a talking clip.

This page is the silent-versus-soundtrack pick at that sticker. It is not a director-versus-volume essay. If the brief needs a voice in the generate, this is the wrong family at any length. If the brief needs a 5s or 10s plate you can steer, 50 credits is the listed price of that plate.

Fifty credits, no soundtrack

The published row: Runway Gen-4.5 video generation with cinematic motion and precise camera controls. audio: false. credits: 50. Durations: 5s, 10s. Cats: T2V and I2V. req: false.

Read those fields in order. False audio is the one teams skip because the landing-page copy says cinematic. Cinematic is a look. It is not a mix. Prompting “a narrator says the price” may paint a mouth or burn letters into the frame. It will not return a VO track you can recut.

What you do instead, if the cut must speak:

• Generate the silent plate here (or do not).
• Generate speech on Gemini 3.1 Flash TTS at 4 credits or Cartesia Sonic 3.5 at 5.
• Lay the voice. Caption the result. Burned-in captions still matter on mute; the missing native track is not an excuse to skip type.

What you do instead, if the cut must speak in the generate:

• Leave this row. Veo 3.1 is 40 credits, T2V, native audio, 4K, 4 / 6 / 8s.
• Flux 3 Text to Video is 9 credits, T2V, native audio, 5–20s.
• LTX 2.3 Text to Video Fast is 4 credits, T2V, optional native audio, 6–20s, 4K listing.
• Seedance 1.5 Pro Text to Video is 6 credits, T2V, native audio, 1080p, 4–12s.

Those stickers are not “cheaper Runway.” They are different flags. 50 credits of silent control is more expensive than 4–9 credits of native-audio T2V. Pay 50 when the camera path is the deliverable. Do not pay 50 because the word Runway felt like a soundtrack.

Do not convert 50 credits to USD. Do not invent a per-second rate the published card does not list. The row is 50 credits, 5s or 10s.

Five seconds or ten

Two durations. Not 4. Not 6. Not 8. Not 15. A seven-second idea buys 10. A four-second idea buys 5 and you live with it, or you buy 10 and cut. Trimming does not refund.

That staircase is the same shape as Hailuo’s 6s / 10s, on a different family, at a different sticker, with T2V available and audio still false. If you needed 8 seconds of talking cinematic, you wanted Veo’s 8s chip, not a 10-second silent Runway you will talk over.

I2V on this row does not require an image on the published req flag — unlike Hailuo Pro / Fast, which do. You may attach a still. You may also prompt. That is the T2V + I2V pair. It is not a reason to skip the still when the SKU must hold. A pack that has to stay lettered still wants a plate. Control without a still is camera-on-a-guess.

T2V without a still is allowed. Use it when the shot is a room, a sky, a move you can name in a sentence. “Dolly in on the label” without a label still is how you get a cousin bottle. Attach the still if the label is the job.

Five seconds is the default until the move actually needs ten. Ten seconds of an orbit that finished at four is six seconds of drift you will cut and still have paid for. Start at 5s. Promote to 10s when the camera has somewhere to go.

The AI video generator will not add a 7s chip because the brief said “about seven.” Confirm 5 or 10.

Control is the reason to pay

Fifty credits is a lot next to Kling 3 Turbo at 12 (silent T2V, 3–15s, 1080p) and Hailuo Fast at 28 (silent I2V, 6s / 10s). The difference you are paying for is not audio — all three are silent or unset — and not length. It is the kind of brief the row is for: a move you could hand a camera operator.

In:

• Dolly in on the label, stop, hold.
• Orbit 90 degrees, keep the bottle planted.
• Handheld walk-up, no drift on the face.
• One object removed, shadows still legal.

Out:

• “Cinematic product video, make it pop.” That is a one-click brief in expensive clothes. Turbo or Fast will guess as well, cheaper.
• Thirty hook variants. That is a volume grid. 50 × 30 is not a test; it is a burn.
• A talking presenter. Audio false. Wrong flag.

Image-to-video here is still silent 5s or 10s at 50. Midjourney V7 Image to Video is 110, silent, 720p, image required, no duration chips. Use MJ I2V when the still is already V7 and 720p is enough. Use Runway I2V when you have a still or a sentence, you need 5s or 10s, and the camera path is the point. Do not run both on the first still.

Text-to-video here is still silent. If the sentence is the whole brief and you needed sound, Flux 3 or LTX Fast or Seedance 1.5 Pro T2V undercut 50 and turn audio on. If the sentence names a trajectory and you will mix later, 50 can be right.

Caption the keeper. Control does not typeset. Add captions to a video after picture lock. Place the line in the safe area. A precise dolly whose hook sits under TikTok chrome is still an unreadable cut.

Native-audio jobs live on other rows

A routing table for this sticker:

Job	Row	Why not Gen-4.5
5s / 10s silent camera move	Gen-4.5, 50cr	This is the job
Silent pack twitch, already have a still	Hailuo Fast, 28cr	Cheaper I2V, 6s / 10s
Silent T2V draft, 3–15s	Kling 3 Turbo, 12cr	Cheaper, longer, still silent
V7 still, 720p, silent	MJ I2V, 110cr	Style lock, not camera chips
Talking 4K, 4 / 6 / 8s	Veo 3.1, 40cr	Native audio, different length
Talking T2V, 5–20s	Flux 3, 9cr	Native audio, cheaper
Optional native audio, 6–20s, 4K listing	LTX 2.3 Fast, 4cr	Soundtrack available
Cheap native 1080p, 4–12s	Seedance 1.5 Pro T2V, 6cr	Audio true

Gen-4.5 is the first row. Everything else is a different flag or a different sticker. Mixing them in one sentence — “Runway, but make it talk like Veo, at Turbo price” — is how 50 credits buy a silent cousin of a brief that wanted 6 credits of Seedance.

World consistency, motion brush, and the rest of Runway’s own surface are vendor vocabulary. On Versely you confirm a catalog row: 50 credits, 5s or 10s, T2V + I2V, audio false. If the shot needs those durations and that control, take it. If the shot needs a voice in the file, take a row whose audio field is true.

Sora’s API shutdown (OpenAI Help Centre, September 2026: API shuts 24 September 2026) does not make Gen-4.5 a Sora replacement. Sora 2 T2V on this catalog was native audio at 720p. Gen-4.5 is silent. Replacing a talking 720p leftover with a silent 50-credit plate is a mix change, not a migration. Put talking leftovers on a native-audio row.

If the next shot is a specified 5- or 10-second move and you will mix later, run Runway Gen-4.5 from Versely. If the next shot talks in-camera, do not.

FAQ

Does Runway Gen-4.5 include native audio at 50 credits?

No. Audio is false. 50 credits, 5s or 10s, T2V and I2V. Add TTS and captions after, or pick a row whose audio flag is true.

Is 50 credits per second?

Not on the published card. The catalog sticker is 50 credits, durations 5s and 10s. Do not invent a per-second rate.

When is 50 credits the wrong silent row?

When you only needed a cheap pack twitch (Hailuo Fast, 28) or a silent T2V draft (Kling 3 Turbo, 12), or when you needed a voice in the generate (Veo, Flux 3, LTX Fast, Seedance 1.5 Pro T2V). Pay 50 for a specified camera move, not for the word cinematic.