Wan 2.7: pick T2V, I2V, or R2V
Wan 2.7 T2V and I2V are 16 credits; R2V is 20. Audio is null on these video rows. Need native audio? Use Wan 2.6 T2V, Seedance, or Veo.
Wan 2.7 ships three published video generate rows: text-to-video and image-to-video at 16 credits, reference-to-video at 20. Pick by input. Audio is null on these 2.7 video rows in the Versely catalog. If you need native audio, use Wan V2.6 Text to Video, Seedance, or Veo — not a hope that 2.7 “probably has sound.”
Alibaba’s Model Studio cards describe wan2.7-t2v with native audio on Alibaba’s side. That is Alibaba’s surface. The Versely catalog fields for wan-2-7-text-to-video, wan-2-7-image-to-video, and wan-2-7-reference-to-video list audio as null. This page follows the catalog. Do not mix 2.6’s true flag with 2.7’s null.
Wan 3.0 is a separate family, and its cards list native audio. Wan 3.0 Text to Video is 6 credits (30% off the 8-credit price), audio true, 2–15 seconds. Wan 3.0 Prime Text to Video is 11 credits. The three rows below are the 2.7 family.
Three rows, three inputs
Wan 2.7 Text to Video — 16 credits, 1080p, 2–15 seconds, text-to-video, audio null. Prompt in, picture out. No still required. Use it when there is no approved frame and you are exploring motion at 1080p. The catalog description mentions strong motion consistency, optional custom audio input, and prompt rewriting. Optional custom audio input is not the same as catalog native audio true. If you need speech born with the picture as a published native-audio job, this is the wrong flag.
Wan 2.7 Image to Video — 16 credits, 1080p, 2–15 seconds, image-to-video, requires a still, audio null. The catalog description: animates images with three modes — first-frame to video, first-and-last-frame interpolation, or video continuation, with optional driving audio. The still is the contract. Use it when a JPEG already passed review and you want that frame to move. Optional driving audio is an input, not a native-audio true flag on the row.
Wan 2.7 Reference to Video — 20 credits, 1080p, 2–10 seconds, reference-to-video, requires refs, audio null. Description: generate from up to 5 reference images and/or videos, with optional first frame and voice timbre control. This is the stack job. Ceiling is 10s, not 15s. The extra 4 credits over T2V/I2V buy the multi-ref door, not a longer clip and not native audio true.
Pick with a one-line test:
- No file in hand → T2V.
- One approved still (or first+last) → I2V.
- A pile of pack/person/set refs → R2V, and accept the 10s cap.
Do not run T2V “plus a mention of the product name” when you have a pack photo. That is how lettering dies. Do not run R2V with one still because the slug sounds more serious. One still is I2V.
Edit is a fourth object: Wan 2.7 Video Edit at 20 credits. That is not a generate. It is the next page.
Duration and resolution are not the pick
T2V and I2V share 2–15s at 1080p. R2V is 2–10s at 1080p. All three are 1080p on the card. You do not pick among them for sharpness. You pick for input.
Two seconds is a legal tick. It is also a blink. If the brief is a 10-second turn, say 10s. If the brief is a 15-second T2V and you actually have five refs, you cannot stretch R2V to 15s. Cut the action to 10s or drop back to I2V with one still and lose the extra refs.
Compare the ceilings to neighbours so you do not use Wan 2.7 for a length it does not have:
- Veo 3.1: 4 / 6 / 8s, 4K, native audio true, 64 credits.
- Seedance 2.0: 4–15s, 1080p, native audio true, 44 credits for T2V, 97 for R2V.
- Seedance 2.5: 4–30s, 720p, native audio true, 34 credits.
- Flux 3 T2V: 5–20s, native audio true, 34 credits.
- Kling 3 Turbo: 3–15s, 1080p, audio false, 22 credits.
Wan 2.7 T2V at 16 credits sits just under Turbo’s 22 on price and matches it on length, with audio null instead of false, and 2s as a floor instead of 3s. That is not a reason to treat them as twins. Turbo is the silent draft row we document as audio false. Wan 2.7 T2V is the 2.7 prompt row whose audio field is unset.
R2V at 20 credits is cheaper than Seedance 2.0’s 97 for reference-to-video, shorter (10s vs 15s), 1080p like 2.0, and without native audio true. If the SKU must hold and speak, Seedance. If the SKU must hold for 8 seconds and you will score it later, Wan 2.7 R2V is in the running.
If you need native audio, leave these three rows
The catalog is the rule. Audio null on 2.7 T2V, I2V, and R2V means this page will not promise a talking take.
Wan V2.6 Text to Video — 40 credits, 1080p, 5 / 10 / 15s, audio true. A higher sticker than 2.7 T2V’s 16, native audio on, three ticks not a 2–15 slider. Use 2.6 when the family should stay Wan and the file must include sound. Wan 3.0, above, also lists audio true, from 6 credits. Do not assume 2.7 inherited 2.6’s flag.
Seedance 2.0 — 44 credits for T2V (97 for R2V), 4–15s, 1080p, native audio. Use when identity and sound both matter.
Veo 3.1 — 64 credits, 4 / 6 / 8s, 4K, native audio. Use when the job is short dialogue at 4K.
Flux 3 T2V — 34 credits, 5–20s, native audio. Use when the job is a 20-second prompt with sound and no still.
If you attach a driving audio file to Wan 2.7 I2V because the description allows optional driving audio, you are feeding a track, not flipping the catalog audio field to true. Verify the output. Do not write “native audio” in the edit decision list.
Captioning is still required on mute feeds. Burned-in captions do not care that Wan 2.7’s audio field is null. Add captions to video after you have a file.
A routing order that does not thrash
- Decide whether native audio is required. If yes, skip 2.7 video generates.
- If no, count inputs: zero, one, or many.
- T2V / I2V / R2V accordingly. Respect R2V’s 10s cap.
- Inspect identity before a second generate. A miss on T2V is not fixed by R2V with the miss as a “ref.”
- If the take is almost right, that is Wan 2.7 edit, not a fourth generate.
Run 2.7 from the AI video generator with the slug that matches the input. Do not ask the agent to “use Wan” without naming T2V, I2V, or R2V.
When the input is decided, run that row from Versely.
FAQ
Does Wan 2.7 text-to-video have native audio on Versely?
The catalog field is null. Do not claim native audio on wan-2-7-text-to-video, wan-2-7-image-to-video, or wan-2-7-reference-to-video. For a Wan family row with audio true, use Wan V2.6 Text to Video (40 credits, 5 / 10 / 15s, 1080p).
Why is R2V 20 credits and capped at 10s?
That is the published card: 20 credits, 2–10s, 1080p, up to five refs. T2V and I2V are 16 credits and 2–15s. You pay 4 credits for the stack door and you give back 5 seconds of ceiling. If you need 12–15s with refs and native audio, Seedance 2.0 is the other family.
Is Wan 3.0 the upgrade path?
Partly. Wan 3.0 is in the catalog: Wan 3.0 Text to Video is 6 credits (30% off the 8-credit price), audio true, 2–15 seconds, and Wan 3.0 Prime Text to Video is 11 credits with the same audio and length flags. Both list audio where 2.7 lists null, and neither card lists a max resolution, where 2.7 lists 1080p. If the file needs sound, check 3.0 first. If you need 2.7’s 1080p listing, its R2V stack door, or its edit row, stay on 2.7.