Veo 3.1 T2V vs first-last

Veo 3.1 is 64 credits, text-to-video, native audio true, 4K, 4 / 6 / 8 seconds, no image required. VEO First Last Frame is 64 credits, image-to-video, native audio true, 4K, 4 / 6 / 8 seconds, image required. Same sticker. Same microphone. Same ceiling. Same length menu. The fork is the input: a sentence versus two frames. Both bill 16 credits a second silent and 32 with sound, so the 64 is the 4s silent floor and an 8s take with sound is 256.

This is not first-last versus R2V. R2V is 128 credits, 8s only, a stack. This page is the two 64-credit Veo doors. Google Flow will run both ideas as “Veo.” Flow is a surface. The catalog shows which gate. Keep Veo. Do not live in Flow.

Both cards mark audio true. Write whether anyone speaks. A surprise monologue is waste on either door. SKU work usually wants Foley and no voice. Say “no speech.”

T2V is the sentence at 64

Use Veo 3.1 when there is no plate that must be frame one. The model will invent the room, the pack, the hands. If the SKU must match a PDP photo, stop. You are on first-last (two frames) or on R2V (stack) or on another family’s I2V.

Duration: 4s, 6s, 8s. Draft at 4s. Step to 6s or 8s when the motion needs the beats. There is no 10s on this slug. Gemini Omni Video is 29 credits, 4 / 6 / 8 / 10 seconds, 4K, audio null, if 10s at 4K matters more than Veo’s microphone. Different flag. Different sticker.

Resolution: 4K listed. That is the official 4K tier on this card. A bad sentence at 4K is a sharper wrong product. Approve the idea at 4s before you spend 8s of 4K on a cousin.

When T2V is the pick:

• You have a cinematic sentence and no locked stills.
• 4, 6, or 8 seconds at 4K with native audio is the file.
• Inventing the pack is acceptable, or there is no pack.

When T2V is the wrong pick:

• You have a start frame and an end frame that already are the shot. First-last at the same 64 will interpolate them.
• You have a reference stack and a new scene. That is R2V at 128, 8s.
• You needed 10s or 30s. Veo T2V stops at 8s. Grok Imagine Video starts at 24 credits, 720p, up to 30s, native audio.

Sixty-four credits is the catalog sticker. Do not call T2V “cheaper than first-last.” It is the same 64.

First-last is two frames at the same 64

Bring stills you would print. Frame one is the upload. Frame last is the upload. The path between them is the generate. The kitchen in the first photo should survive if the last photo is the same kitchen. If you need a new kitchen, you wanted R2V or T2V, not a morph that has to reconcile two rooms.

Same 4 / 6 / 8. Same 4K. Same native audio. Same 64. You are not paying extra for the stills. You are spending the Veo 64 on a different gate. The stills are the savings: you do not have to describe the pack in prose.

When first-last is the pick:

• Closed-to-open, empty-to-filled, before-to-after, on plates you already trust.
• 4K dialogue or 4K Foley in 4, 6, or 8 seconds is acceptable.
• Identity is already in those two frames.

When first-last is the wrong pick:

• You have only a sentence. The gate wants an image. Use T2V.
• You have one still and no destination frame. Single-still I2V is a different family, or you duplicate the still and pretend — that is a bad interpolate. Don’t.
• You needed a stack in a new location. R2V at 128.

Flux 3 First Last Frame to Video is 34 credits, native audio true, 5–20 seconds, if 64 is too much Veo for a morph and 4K is not required. Pixverse Transition is 12 credits (32 at 1080p), edit-video, 1080p, audio null, 5 / 8 / 10s — well under 64, no microphone, no 4K. If you needed Veo’s 4K talking morph, stay. If you needed a cheap mute interpolate, leave.

Same 64, pick the gate before you prompt

Do not run T2V to “see the vibe” and then first-last to lock the pack. The T2V cousin is not frame one. If the pack exists, start on first-last. If it does not, start on T2V, or shoot the stills.

Do not run both at 64 + 64 as a default stack. Name the stack if you must: T2V for a wide, first-last for the cap morph. If you cannot name it, pick one.

VEO 3.1 extend is 112 credits (224 with sound), extend-video, native audio, 7s, 720p. That is a third Veo gate, a hop, a 720p ceiling. It is not this comparison. If you needed more than 8s of Veo, extend is the other gate — not a longer T2V chip, and not a first-last chip.

The AI video generator is the launcher. Captions still belong on talking 4K. Mute autoplay hides native audio. Captions for silent autoplay is the typesetting. The AI caption generator burns the line.

Article 50 still applies to photoreal people. Sixty-four credits does not exempt a generated face. Label at first exposure for EU audiences since 2 August 2026. The Article 50 disclosure template is the copy. YouTube’s 27 May 2026 Shorts overlay labels photoreal AI. Labels do not turn a sentence into two frames.

Sora 2 T2V was 32 credits, native audio, 720p. Sora 2 Pro storyboard was I2V, native audio, 10 / 15 / 25s, 240 credits. The Sora API shuts 24 September 2026 (OpenAI Help Centre, September 2026). A leftover Sora sentence at 720p is closer to Grok at 24 than to Veo T2V at 64 unless you need 4K in 8s. A leftover two-frame brief is first-last at 64 if 8s 4K is enough.

UGC that is a real product still belongs on first-last. UGC that is a premise with no photo belongs on T2V. Veo will not move the still through the sentence door.

What the shared 64 is not

64 is not 10s. 64 is not R2V. 64 is not Omni’s 29-credit null card. 64 is Veo’s T2V or Veo’s first-last: native audio, 4K, 4 / 6 / 8 seconds. The only decision on this page is whether you have the frames.

Run the 64-credit gate you actually have in Versely.

FAQ

Is Veo first-last more expensive than Veo 3.1 T2V?

No. Both published stickers are 64 credits. You pick by input: sentence versus two frames.

Do both rows have native audio and 4K?

Yes. audio true and 4K on both cards. Same length menu: 4s, 6s, 8s.

Can I use one still on first-last?

The job is first and last frames. One still is not two ends. Use T2V, a single-still I2V family, or bring a real destination frame.