Google's general-purpose Gemini image model. Open it when a Visxy library prompt already describes the picture and you want a finished frame, not a thumbnail.

The fast lane of Google's Nano Banana family. Built for trying Visxy's long 9:16 prompts — not for print-size files or long edit chains.

The high-fidelity member of the family. Use it when a Visxy prompt has to survive a close look — type, hands, labels — not when you are still shopping looks.

OpenAI's current image model. Built for a long Visxy library prompt — layout, type, and what to keep — not a five-word vibe.

ByteDance's current Seedream stills model. For a Visxy library prompt that has to look designed — type, layout, a 9:16 poster — not a five-word vibe.

ByteDance's cheap Seedream 5 stills SKU. Try a Visxy 9:16 poster here — not Seedream 5 Pro, and not Seedance.

Black Forest Labs' production image model. Photoreal product and materials — not the Flux 3 video row.

Black Forest Labs' cheaper Flux 2 stills SKU — Flex under a short catalog name. Same frames as Flux 2 Pro, not the Pro row and not Flux 3.

Ideogram's type-first image model. Use it when the Visxy prompt is a poster, a label, or a word that has to read.

xAI's current Grok stills model. A Visxy 9:16 library prompt — not the unversioned Imagine row, and not a video model.

xAI's unversioned Grok stills model. Fast 9:16 tries — not Grok Imagine 2.0, and not a video row.

Kuaishou's cinematic clip model. One subject and one camera move that stay coherent — not the five-second Turbo try.

Kuaishou's Omni Kling beside Kling 3.0. Same coherent-move job, with a size picker — not Turbo, not 2.6, and not a multi-file studio.

The short Kling lane next to Kling 3.0. A five-second try — not a setting on the 3.0 page.

Kuaishou's previous Kling. Five or ten seconds, optional sound — not 3.0, not Omni, and not the five-second Turbo row.

Google's current Veo quality tier. Short cinematic clips with sound, 9:16 from the video library — not Fast, not Lite.

ByteDance's long-take video model. Use it when a Visxy video prompt has a sequence to hold — not when you only need a few seconds of motion.

The short draft lane next to Seedance 2.5. Try the shot here; move a keeper to 2.5 only when the clip has to run past a quarter-minute.

MiniMax's current Hailuo video model. For a Visxy 9:16 clip with one clear action — not a stills generator.

Black Forest Labs' video model. A clip with sound in the same pass — not the Flux 2 / Flux 2 Pro stills rows.

Alibaba's current Wan video model. Smooth motion from a Visxy 9:16 prompt, or a still with an optional last frame.

PixVerse V6 under a short catalog name. Short-form motion, audio off until you turn it on — not V5.

xAI's unversioned Grok video model. A longer clip than 1.5 — not the stills rows, and not a five-second sting.

xAI's versioned Grok video model. A short clip from a Visxy prompt or one still — not the unversioned Video row.

Luma's current Ray. A short directed clip — five or ten seconds — not a 16-keyframe studio and not Ray 2.