Everygen AI Video Generator: Models & Prompts

Your first generation

Write a one-sentence brief. Name the subject, action and intended use. A five-second product detail is a clearer starting point than a complete advertisement.
Open the official Everygen site. Sign in there, choose the video tool and select the model. This guide never needs your account credentials.
Set duration, resolution and aspect ratio. Use 9:16 for a vertical draft or 16:9 for a horizontal draft. Confirm that your selected model and plan support them.
Attach any references, then review the credit quote. Check rights to uploaded images and get consent for identifiable people.
Generate once and inspect. Check the first frame, last frame, identity, geometry, text and sound. Save the prompt and settings alongside the output.
Iterate with a fixed budget. Change the camera move or action, rather than rewriting everything. Stop when the attempt budget is spent.

Four model starting points

Seedance 2.5

4–30 seconds · up to 1080pSound supported

Try it for: A deliberately paced product shot or a scene that needs more than 15 seconds. Start with a short draft before committing to a longer take.

Suggested starting settings: 5-second draft · 720p · 16:9. These are an example brief, not an output we generated.

A small ceramic teapot sits beside a rain-streaked window. One slow push-in as a hand lifts the lid and steam rises. Soft grey daylight, quiet rain and ceramic sounds. Keep the teapot shape consistent. No text or music.

Official model specifications and output examples ↗. Examples on that page belong to Everygen, not this site.

Kling 4.0

3–30 seconds · up to 1080pSound supported

Try it for: A shot anchored to a reference image. Upload media you have permission to use and specify which details must remain fixed.

Suggested starting settings: 5-second draft · 720p · 16:9. These are an example brief, not an output we generated.

Using the attached image as the product reference, show the same olive-green backpack on a wooden bench. A hand opens the main zip; the camera slides slowly from left to right. Keep stitching, proportions and color unchanged. Natural room sound, no text.

Official model specifications and output examples ↗. Examples on that page belong to Everygen, not this site.

MiniMax H3 Max

5–15 seconds · up to 720pSound supported

Try it for: A short moment where the action and its sound matter together. H3 Max Turbo is a separate option with different limits, including up to 1080p.

Suggested starting settings: 5-second draft · 720p · 16:9. These are an example brief, not an output we generated.

A baker taps a loaf of bread, slices it once, and places a slice on a board. One static close shot, warm kitchen light. Hear the crisp crust and the knife touching the board. No speech, music, captions or extra hands.

Official model specifications and output examples ↗. Examples on that page belong to Everygen, not this site.

Grok Imagine

1–15 seconds · up to 720pSound supported

Try it for: A compact stylized idea. Keep the action readable and evaluate visual consistency before attempting more complicated motion.

Suggested starting settings: 5-second draft · 720p · 16:9. These are an example brief, not an output we generated.

A tiny astronaut waters a single sunflower inside a round greenhouse on the moon. One gentle camera orbit, soft reflected Earth light. The flower sways slightly. Quiet water drops and suit movement. No captions or additional characters.

Official model specifications and output examples ↗. Examples on that page belong to Everygen, not this site.

Build an editable brief

THE PROMPT WORKBENCH

Choose a model

Choose a scene

Model: Seedance 2.5. 5 seconds, 720p, 9:16. A matte green travel bottle on a sandstone plinth. A single slow orbit reveals the brushed metal lid. Soft window light from the left, quiet studio ambience. Keep the bottle shape and label unchanged. No captions, extra objects or music.

What to check in the result

A successful render is not automatically a usable clip. Watch it once without sound, then listen without looking at the picture. This separates motion problems from audio problems.

Check	What to look for	What to change next
Identity	Does the subject change shape, color or face?	Simplify the action and reuse a clear reference.
Motion	Do objects intersect or hands multiply?	Use one action and a slower camera move.
Sound	Does speech match the intended words?	Shorten the line; separate dialogue from ambient sound.
Brand details	Is a label distorted or invented?	Avoid readable text in generation; add exact typography in editing.
Delivery	Is the framing right for the destination?	Confirm aspect ratio and resolution before generating again.

Do not use a polished sample from a provider as a promise of the result you will get. The official model pages above offer real provider examples; our prompts are original and have not been run.

Keep retries in the budget

Record every attempt, including rejects. If a clip takes four attempts, its effective credit cost is roughly four times the cost shown for a single render, before editing and export costs. Use our cost calculator to compare that with your plan.

For a useful model comparison, keep the brief, duration, resolution and reference media constant. Compare usable outputs and total credits rather than one selected frame.

Sources & verification

Primary sources checked November 5, 2026. Product details may change.

Specifications are provider statements, not an independent benchmark. No paid generation, speed test or quality ranking was performed for this guide.

Your first generation

Four model starting points

Seedance 2.5

Kling 4.0

MiniMax H3 Max

Grok Imagine

Build an editable brief

What to check in the result

Keep retries in the budget

Sources & verification

Connect your AI assistant

Understand the real cost

Decide if it fits your workflow