Flux 2 Max is a stills row

Flux 2 Max is a stills row. It does not generate video. Flux 2 Max is published as text-to-image, 6 credits, content type image, no duration list, audio empty. The catalog line is maximum quality image generation for ultra-detailed photorealistic outputs. Prompting “slow push in, 5 seconds” will not produce 5 seconds. It will produce a detailed still of a camera that never rolled. Lock the plate here. Then pick a motion model. Do not treat Max as a quiet Veo.

This page is the category error: Flux 2 Max is not video, including not Flux 3 video. It is the published stills SKU. Animate after approval.

What the card actually is

Read the fields.

• ctype: image
• cats: text-to-image
• credits: 7
• durs: empty
• audio: null
• requires_image: false
• provider: Flux

Text is enough. That is the point of a stills row: cheap discovery of the face and the SKU before anyone spends 64 credits or more on motion. Six is more than a 1-credit sketch row. It is still a still. The extra credits are the photoreal bet: ultra-detailed outputs when the frame has to survive a zoom on a label or a face.

Aspects on typical Flux stills runs include the usual delivery ratios (16:9, 9:16, 1:1, and the 4:3 family). Crop for the platform you will animate later. Do not promise a client a 4K Flux 2 Max export from a blank max-output field. If the insertion order said 4K stills, confirm what the card lists before you say yes.

The Flux roster also has Flex, Klein, Flash, Kontext, Pro, Schnell. Max is the photoreal still at 6 credits. Flex is a 4-credit flexible still/edit cousin. None of those are video because a sibling named Flux 3 exists. Flux 3 Text to Video and Flux 3 First Last Frame to Video are different slugs. Opening Max because you saw “Flux” and wanted motion is the error this page exists to stop.

Text to image is the launcher. Confirm image, 6 credits, no duration menu. If you see a duration menu, you are not on Max.

Why people treat it as video anyway

Three habits:

1. The prompt is written like a shot list. “Dolly in, 4K, 8 seconds, native audio, product turn.” Max will try to draw a dolly. You will get a still that looks like a frame from a dolly. Teams then say “Flux video is weak.” They never left stills.

2. The family name moved. Flux 3 video rows landed on the same roster. People grep “Flux,” sort by familiarity, pick Max because they used it for hero stills last quarter. The 3-line is the video. The 2 Max line is the still.

3. Cost looks like a cheap clip. 6 credits versus Veo’s 64 feels like a deal on a “quick video.” It is a deal on a still. The clip still has to be bought. Total cost is 6 plus the I2V row, not 7 instead of the I2V row.

The correct sequence:

1. Generate or edit the still on Max until you would print it.
2. If one object is wrong, use an edit model. Re-rolling a 6-credit photoreal still to fix a zipper is how Max gets used as a retoucher.
3. Send the approved file to image-to-video. Hailuo 2.3 Fast is 12 credits at 6s or 23 at 10s when cheap short motion is enough. Veo first-last is 64 credits when two stills are endpoints. Omni Video is 29 credits when 4–10s Omni I2V is the look. Seedance 2.5 is the wrong next step if you needed R2V; 2.5’s published card is text-to-video.

Keep the still. The next variant should start from this file, not from a new paragraph that almost describes it.

What Max is for, and what it is not

Max is for:

• Photoreal hero stills you will later animate.
• Pack shots that must survive a zoom on type and material.
• A lock you can reject in 6 credits instead of discovering the jacket is wrong at 64.

Max is not for:

• Any duration. Empty durs is the contract.
• Native audio. There is no video, so there is no band.
• First-last interpolation. That is a video category. Flux 3 has a first-last video slug if you meant that family. Veo has VEO First Last Frame.
• Reference-to-video stacks. Three stills into Veo R2V is 64 credits of recast, after Max (or another stills row) made the stills.
• Captions. Type in a still is pixels. Type on a video is an editor. Add captions to video after motion exists.

Versus other stills rows, briefly: GPT Image 2 Text to Image is 9 credits (was 17), 4K listed, OpenAI, prompt adherence. Nano Banana 2 is 4 credits (was 7), Google, T2I + I2I + edit. Max is the Flux photoreal 6-credit lock. Pick Max when the Flux look and the photoreal bet are the brief, not because it is “the video model we already know.”

Article 50 still applies to a realistic still of a real person if you publish it as a synthetic in scope. Video disclosure is the common case; do not assume a still is exempt without reading the Article 50 page.

A week that does not pretend Max moves

Monday: 10 Max stills of the SKU. Kill 8.

Tuesday: one edit pass on the keeper if a button is wrong. Not 10 new Max rolls.

Wednesday: I2V on the keeper at 6s. If the motion is the product and 6s is enough, Hailuo 2.3 Fast. If the stills were a cast for a new scene, you never wanted I2V — Veo R2V, and Max was only how you built still two.

Thursday: captions, crop, disclose.

Friday: do not open Max to “see it in motion.” Open a video row.

If someone on the team says “just Flux it,” ask still or video. If still, Flux 2 Max. If video, name a video slug. The AI video generator will not run Max as a clip. The duration menu will be empty because the content type is image.

Lock the plate at 6 credits, then sign up and pick the motion row on purpose. Max is the still. It stays the still.

FAQ

Can Flux 2 Max make a 5-second clip?

No. The published card is text-to-image, 6 credits, no durations. For Flux-family video, use a Flux 3 video slug, not Max.

Is Max the same as “lock this still before you buy motion”?

That workflow is how you use a stills row. This page is the category fact: Max is not video. Do not skip the still, and do not skip the second model either.

Why 6 credits for one picture?

It is the photoreal still, not the cheapest sketch. Volume thumbnails belong on a 1-credit-class row. Spend 7 when the still is the product you will animate or print.