Luma Ray 2 720p: 40 credits for a 5s clip

Luma Ray 2 720p starts at 40 credits (a full job runs 40 to 80 depending on clip length), text-to-video, 720p, 5s or 9s, audio null, no image required. Forty credits is the listed floor. It is a lot of sticker for a mute-or-unset 720p file. Grok Imagine Video starts at 24 credits, native audio true, 720p, 6 / 10 / 15 / 20 / 30 seconds. If you opened Ray 2 because you remember a demo, read the SKU.

This is not a “pick the row not the brand” restack. That guide already said 720p is the product. This page is the job pick: 40-credit 5s/9s null versus cheaper 720p talking Grok, versus 1080p silent neighbors, versus 4K talking Veo. Forty has to win on purpose.

The card: category text-to-video, durations 5s and 9s, resolution 720p, from 40 credits, audio unset. Description: Luma Ray 2 video generation in 720p quality. Quality here is the tier name, not a compliment. There is no 1080p switch on this slug.

720p is the ceiling, 40 is the floor

Max output is 720p. If the delivery spec is 1080p or 4K, this generate is look-dev or the wrong row. Upscaling a 720p take on ByteDance Upscaler Video (from 3 credits) is a finish pass, not the same as picking a 1080p model. Budget the upscale only after picture lock. Do not upscale drafts.

Duration: 5s or 9s. There is no 6s, no 8s, no 10s, no 30s. Grok’s shortest talking 720p chip is 6s. Ray 2’s shortest is 5s. That one-second difference is not a reason to pay Ray 2’s price. 9s is Ray 2’s long hold. Grok sells 10s from 24 with a microphone. If 9s mute is truly the file, name why Grok’s 10s talking 720p is wrong.

Audio null: do not write a musical and expect a native bed. Do not write a presenter line. Plan a pass: generate the picture, attach audio in the editor, or pick a true row. Captions for silent autoplay is the typesetting if you stay.

When Ray 2 720p is the pick:

• You want this family’s motion at 5s or 9s, and you have seen it beat Grok on that named motion.
• 720p is the master, or 720p is look-dev you accept at 40 or more.
• Null audio is acceptable.
• You have a sentence, not a still that must be frame one.

When Ray 2 720p is the wrong pick:

• You needed native audio at 720p. Grok starts at 24.
• You needed 15, 20, or 30 seconds at 720p. Grok sells those chips. Ray 2 does not.
• You needed 1080p as the generate. Pixverse V6 Text to Video starts at 10 credits, audio null, 5s / 8s, 1080p. Hailuo 02 starts at 32 credits, audio false, 1080p, no duration chips.
• You needed 4K dialogue. Veo 3.1 starts at 64 credits, native audio, 4 / 6 / 8 seconds.

Forty credits is the catalog floor. Versely bills in credits. Do not call 40 “Luma pricing so it must be the hero.” The hero is the file that matches the chips.

Forty versus twenty-four at the same 720p

Grok Imagine Video: audio true, 720p, 6–30s, from 24 credits, T2V. Ray 2 720p: audio null, 720p, 5s / 9s, from 40 credits, T2V. Same resolution class. Different flag, different length menu, different sticker. A talking 6-to-30-second menu from 24 versus a mute 5s-or-9s menu from 40.

Pay for Ray 2 when Luma’s motion is the brief and Grok’s motion is wrong in a way you can name. Do not pay 40 because the brand is Luma. Do not pay it as a “premium 720p.” Premium would need a flag or a ceiling the card does not have.

Flux 3 Text to Video starts at 34 credits, native audio true, 5–20 seconds, resolution unset on the snapshot, description up to 1080p. That is another talking generate, below Ray 2’s floor. Ray 2 is not competing on sticker. It is competing on a look.

Do not run Ray 2 as the first 720p draft. Run Grok (from 24) if 720p talking is acceptable. Run V6 (from 10) if 1080p silent 5s/8s is acceptable. Open Ray 2 when those misses are documented.

No still required. Text-to-video will invent the subject. If the subject is already approved as a still, leave. I2V is a different Luma slug if one exists in your recipe; this SKU is T2V. Do not invent an I2V gate here.

After a Ray 2 720p take

Burn a caption. A Ray 2 take did not come with a mix. The AI caption generator is the typesetting. Burned-in captions are the short-form object. Cartesia Sonic 3.5 is 4 credits per 1,000 characters TTS if you need a line under the plate. A Grok take (from 24 credits) would have come with its own audio, if Grok’s picture was good enough.

If delivery needs 4K, lock picture, then a ByteDance upscale (from 3 credits). Do not regenerate hoping 720p becomes 4K. The res field will not move.

Article 50 still applies to photoreal people. Forty 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 does not care what you paid.

Sora 2 T2V was native audio, 720p, up to 20s. The Sora API shut on 24 September 2026 (OpenAI deprecations). A leftover Sora 720p talking brief is Grok (from 24), not Ray 2 from 40 with a null card. Map the flag. This page is not a Sora essay.

The AI video generator is the launcher. UGC that is a premise at 720p does not get more authentic because the sticker is 40.

What 40 is not

40 is not 1080p. 40 is not native audio. 40 is not 30s. 40 is not a demo you saw in 2024. 40 is Luma Ray 2 720p: T2V, 5s or 9s, audio null. Pay it when that look at those chips is the file. Skip it when talking 720p from 24 credits is the file.

Run the 720p row you can actually defend in Versely.

FAQ

Does Luma Ray 2 720p have native audio?

The published card is audio null. Do not brief in-model speech. Grok Imagine Video is audio true from 24 credits, also 720p.

Why is it 40 credits if Grok starts at 24 at 720p?

Different family, different motion, different flag. Forty is not a 720p tax. It is this SKU’s sticker (40 for a 5-second block, up to 80). Default to Grok’s 24 unless Luma’s motion is the named brief.

Can I get 1080p from this slug?

No. res is 720p. Pick a 1080p generate, or lock a 720p take and upscale on a different row after picture lock.