Replace Sora drafts with Kling Turbo

Cheap silent drafts that used to burn Sora credits belong on Kling 3 Turbo Text to Video: 22 credits, 3–15 seconds, 1080p, audio false. Then add TTS, or promote one keeper to a native-audio hero. Do not claim Turbo has native audio. The catalog field is false.

Sora 2 Text to Video was 32 credits, 720p, native audio, 4 / 8 / 12 / 16 / 20s. Drafting on a native-audio row means you pay for a soundtrack you will throw away. After 24 September 2026 the Sora API is gone per OpenAI’s Help Centre. The draft habit has to land somewhere that still exists and does not pretend to be a hero.

This is not a Sora versus Kling taste piece. It is the silent look-dev row.

What Turbo is, and what audio false means

Kling 3 Turbo T2V: category text-to-video, audio false, durations every second from 3s through 15s, 1080p, 22 credits, discounted on the card. Description: fast text-to-video. That is the whole compliment.

Audio false means the row does not generate a native soundtrack as part of the published job. You will not get Sora-style speech-in-the-take. You will get picture. If you need words, you add them: Gemini 3.1 Flash TTS at 12 credits, another TTS row, or a later native-audio generate on a different model.

That split is the point of a draft:

Spend 22 credits to see whether the camera move, blocking, and prompt even work.
Throw away misses without having paid a 64-credit Veo dialogue bill or a 97-credit Seedance 2.0 reference-to-video bill.
Promote one composition to Veo 3.1, Flux 3 Text to Video, or Seedance 2.0 when the picture is worth sound.

If you write a full spoken paragraph into a Turbo prompt, you are briefing a model that will not say it. Put the line in a caption doc. Put the picture description in the prompt.

Fifteen seconds is the ceiling. Sora’s 16s and 20s drafts do not fit. Cut the draft to 15s or less. You were going to cut it in review anyway.

1080p on Turbo is higher than Sora T2V’s 720p card. Drafts can look “better” than the Sora files you were testing with. Do not confuse a sharper silent plate with a finished talking ad.

How to run a Sora-shaped draft on Turbo

Steal the visual half of the Sora prompt. Camera, subject, motion, duration tick. Drop the dialogue block. Drop “with ambient score.” Those instructions do not bind on audio false.

Pick 3 / 5 / 8 / 10 / 12 / 15 on purpose. Match the length you will actually use in the hero. Drafting 15s for an 8-second Veo hero wastes the end of the clip and teaches you the wrong pace.

Batch the variants. Three framings, one still-unneeded T2V prompt family. Turbo is the row where extras are cheap enough to be honest. Do not “just this once” draft on Veo because the prompt is already written.

Mark keepers as picture-only. A Turbo keeper is a composition. It is not a master with sound. If you need to restyle later, export the frames. If you need speech, that is a new generate or a TTS+lipsync stack.

Caption the silent file if you will test it in a feed. Captions for silent autoplay exist because autoplay is mute. AI caption generator and add captions to video will typeset a line Turbo never spoke. That is legal for a draft. It is not native audio.

Do not use Turbo as the SKU lock. Without a still-contract row, T2V will drift the pack. Draft layout on Turbo; lock identity on Seedance R2V. The SKU replacement page is that job.

Do not use Turbo as the talking-hero replacement. The Veo dialogue page is that job.

After the silent draft: TTS or a native-audio hero

Two promotion paths. Pick one. Do not do both on the same file without a reason.

Path A — keep the Turbo picture, add voice. Generate speech on Gemini 3.1 Flash TTS (12 credits, inline audio tags) or another TTS card. Lay it under the silent clip. If a mouth must match, that is a lipsync row, not Turbo growing a voice. This path is for drafts that became good enough to ship as picture, with words as a layer.

Path B — rerun the keeper prompt on a native-audio row. Flux 3 T2V is 34 credits, 5–20s, native audio. Veo 3.1 is 64 credits, 4 / 6 / 8s, 4K, native audio. Grok Imagine Video is 24 credits, 6 / 10 / 15 / 20 / 30s, 720p, native audio. LTX 2.3 Fast is 20 credits, 6–20s, 4K, native audio. Path B invents a new picture that rhymes with the draft. It does not preserve Turbo’s exact frames.

Path A preserves frames. Path B preserves the idea. Sora used to hide this choice because every generate had native audio. The sunset makes the choice visible. That is useful. Drafts should be cheap and silent. Heroes should be the row that matches dialogue, SKU, or length.

Wan 2.7 Text to Video is 16 credits, 2–15s, 1080p, audio null. It is not Turbo (audio false, 22 credits) and it is not a native-audio hero. Do not park Sora drafts there expecting sound.

Kling O3 Pro T2V is 56 credits, 3–15s, native audio. That is a hero-priced Kling row, not a draft row. Do not “upgrade the draft” from Turbo to O3 Pro unless the brief actually needs O3 Pro.

What not to do on the last Sora days

Do not spend 24 September 2026 generating Sora drafts “while you still can.” Export keepers. Put new look-dev on Turbo. The last-day checklist is stop prompting Sora, not squeeze in more silent-feeling Sora with native audio you will discard.

Do not compare Turbo to Sora on voice. Sora had it. Turbo does not. Compare on whether the shot exists.

Do not write “Kling 3 Turbo with native audio” in a deck. The published card is audio false. Other Kling rows (O3 Pro, V3 4K T2V) have audio true. Mixing those sentences is how a buyer expects a talking Turbo file.

Twenty-two credits is the catalog sticker. It is not USD. It is not a like-for-like swap for Sora’s 32 credits. Sora’s 32 includes native audio at 720p up to 20s. Turbo’s 22 is silent 1080p up to 15s. Different object, different number.

Run the silent draft row from Versely.

FAQ

Does Kling 3 Turbo have native audio?

No. The catalog field is audio false. 22 credits, 3–15s, 1080p, text-to-video. Add TTS or promote a keeper to a native-audio row. Do not brief it as a talking model.

Can Turbo replace 16s and 20s Sora drafts?

Not at those lengths. Turbo’s ceiling is 15s. Cut the draft, or look-dev the long take on Flux 3 T2V (5–20s) or Grok Imagine Video (20s / 30s ticks) if you accept paying for native audio during drafts.

Should every Sora leftover go through Turbo first?

No. Talking heroes go to Veo 3.1. SKU holds go to Seedance 2.0 or Fast R2V. Turbo is for cheap silent composition tests — the work you used to do on Sora because Sora was already open, not because the take needed Sora’s sound.