PixVerse V6 Max API

PixVerse V6 Max is the top-quality version of PixVerse's sixth-generation video model. It keeps what made V6 popular, clips of any length from 1 to 15 seconds at up to 1080p with optional native sound, and aims its upgrades at the places where V6 was weakest: conversations between several characters, motion that has to follow music, and the fine material detail of products, fabric and skin. PixVerse positions it as the quality-first model of its V series, the one to pick when a shot matters more than speed of iteration.

It arrived in late October 2026, seven months after the original V6, and went straight to PixVerse's API partners rather than arriving with a big launch event. On Unicorn API it is available as one model, pixverse-v6-max-image-to-video: you give it a single start image and a prompt, and it animates the scene, with sound if you ask for it.

That makes V6 Max a targeted release rather than a new generation. If V6 was the flexible all-rounder, Max is the version tuned for finish: better prompt following, more realistic materials and lighting, characters and products that hold their look, and dialogue that lands on the right mouth.

PixVerse and the road to V6 Max

PixVerse is the global video brand of AIsphere, a Beijing company founded in 2023 by Wang Changhu, formerly a researcher at Microsoft Research Asia and ByteDance. PixVerse launched internationally at the start of 2024 and by spring 2026 said it served more than 100 million users in 175 countries. A Series C round in March 2026 made the company a unicorn.

The model line has moved quickly. V5 arrived in August 2025, V5.5 in December brought the first native audio and multi-shot output, and V5.6 in January 2026 refined looks, motion and vocals. PixVerse V6 followed on 30 March 2026 and changed the format: any whole-second length from 1 to 15 seconds instead of fixed durations, 1080p at full length, better camera moves and multi-shot clips with sound from a single prompt. Alongside it PixVerse offered a reference model, V6 Fusion, and a first-to-last-frame V6 Transition model.

V6 earned praise for flexibility and camera language, but reviewers found clear faults. The most quoted was dialogue: in one test, two women speaking on screen were given two male, robotic voices. Busy scenes could lose direction, and product-accurate shots took retries. V6 Max reads as PixVerse's answer to that list.

Like every PixVerse model, V6 Max is closed. PixVerse has published no weights, no parameter count and no technical report, and the model is available only through PixVerse's own apps and API and partners such as Unicorn.

What V6 Max changes over V6

Closer prompt following

PixVerse's first claim for Max is that it does more of what the prompt says. In practice that matters most for prompts with several instructions at once: a subject, an action, a camera move, a lighting style and a sound cue. V6 tended to honour some and drop others; Max is pitched as the version that keeps them together.

Materials, lighting and physical motion

Max is described as rendering more realistic materials and lighting, with more natural textures and movement: fabric that responds to a light breeze, a runner whose direction and pace through a turn look physically right. These are the details that separate a clip that reads as footage from one that reads as generated.

Dialogue between several characters

The upgrade most likely to matter to storytellers is dialogue. PixVerse says Max improves lip sync and, crucially, matches each spoken line to the right character, with natural speaking turns, pauses and facial reactions between them. That targets V6's best-known weakness directly and opens the model to short films, conversational ads and character-driven scenes where two or three people talk.

Motion that follows the music

Max is also tuned for music. PixVerse describes tighter synchronisation between the soundtrack and what happens on screen: character moves, visual effects and camera transitions that land on the beat. For music videos, dance clips and performance scenes, that is the difference between footage that sits on top of a track and footage that is cut to it.

Product details and on-screen text

For commercial work PixVerse highlights sharper product detail and more accurate text rendering. Labels, packaging and headline text are among the hardest things for video models to keep legible as the camera moves, and Max is pitched at exactly that use.

Consistency across shots

Finally, Max is described as keeping characters and products more consistent across shots, so the same face, costume or bottle survives a cut or a camera move without changing.

What PixVerse has published

Very little beyond those claims. There is no technical report, no parameter count, no frame rate figure and no benchmark or arena score for V6 Max at launch, and the model is too new for independent reviews. Everything above is PixVerse's description of the model; the only way to know how far the improvements go is to test them on your own material. What is observable is the interface, which on Unicorn is deliberately simple.

PixVerse V6 Max specs on Unicorn

Four fields are required: prompt, image_urls (exactly one URL, used as the first frame), quality and duration. Sound is off by default; set generate_audio_switch to true for music, effects and dialogue generated in the same pass. A fixed seed with the same settings makes results more repeatable, which helps when you change one word of a prompt and want to see only that change.

PixVerse V6 or V6 Max

The two share the 1 to 15 second range, the four resolutions and optional audio, so the choice comes down to what the shot needs.

Pick V6 Max when a clip has people talking, when motion should follow music, when a product or a line of text must stay crisp, or when the material look of a scene, its fabric, glass, skin and light, is the point.

Stay with PixVerse V6 Image To Video when you want its extra switches, such as multi-clip output and effect templates, which the Max model on Unicorn does not expose. For several reference images use V6 Fusion, and for a start-to-end-frame morph use V6 Transition; neither has a Max version.

Where PixVerse V6 Max shines

Conversations. A two-person exchange, an interview set-up or a short drama scene is where Max's line-to-character matching and speaking turns should show most.

Music-led clips. Dance, performance and product reveals timed to a beat play to its music synchronisation.

Product and beauty shots. Glass, metal, liquids and fabric under controlled light, with packaging text that stays legible, suit its focus on materials and detail.

Exact length. Any whole second from 1 to 15 means a clip can match a slot in an edit, a 6-second bumper or a 15-second spot, without trimming.

Repeatable tests. The seed makes it practical to compare prompt changes fairly before committing to a final render at 1080p.

Limits to know before you build on it

One mode on Unicorn. V6 Max here is image-to-video only. There is no text-only mode, no multi-image references and no first-and-last-frame control; for those, the V6 family's other models remain the tools.

No multi-clip switch. V6 Image To Video can be told to cut between shots; the Max model exposes no such toggle. Describe shots in the prompt and treat cuts as a request rather than a guarantee.

1080p is the ceiling. There is no 2K or 4K option.

Unproven claims. With no benchmarks or independent reviews yet, PixVerse's improvements in dialogue and music sync are promises until your own tests confirm them. Multi-character dialogue was V6's weak spot, so check voices and lip sync before publishing.

Literal prompting still wins. Like V6, Max responds best to concrete, visible description. Abstract metaphors and long lists of competing instructions make results less predictable.

Prompting tips for PixVerse V6 Max

A simple structure works well for the V6 family: subject plus one action plus location; one camera movement plus a specific style, lens, lighting or composition; what must stay stable and what should be absent; then the audio. Because the start image already sets the look, spend the prompt on motion, camera and sound. Put dialogue in quotation marks and name who says each line.

A two-person dialogue scene:

The barista slides a cup across the counter and says, "Same as always?" The customer in the green coat smiles and answers, "You know me too well." Gentle push-in, warm morning light, quiet café chatter and the hiss of the espresso machine.

A music-synced product reveal:

The glass bottle rotates slowly on the wet black stone as a band of light sweeps across it on every beat. Water droplets roll down the glass. Slow orbit, macro lens, deep shadows. Keep the label sharp and readable. Minimal electronic beat.

A material and movement study:

The dancer in the silk dress turns once, the fabric flaring and settling as she stops. Locked-off wide shot, soft studio light. No other people in frame. Cello music, footsteps on the wooden floor.

Draft at 360p or 540p with a fixed seed while you adjust wording, then render the keeper at 720p or 1080p.

PixVerse V6 Max against other video models

Kling 4.0 Reference To Video offers clips up to 30 seconds and many more references; choose it for long takes and heavy reference work, and V6 Max for a short, polished shot animated from one image.

Vidu Q4 Image To Video and MiniMax H3 Image To Video are the natural side-by-side tests for animating a still; run the same start frame through each when realism is the brief.

Wan 2.6 Flash Image To Video is another image-to-video model released this month, worth a comparison on the same still before you settle on one.

Getting started

Open PixVerse V6 Max in the studio, upload a start image, choose a length and quality, and switch sound on if the scene needs it. For code, create a key on the API keys page and submit a job to /v1/jobs with model set to pixverse-v6-max-image-to-video, a prompt, one URL in image_urls, a quality and a duration. The API tab on this page has full request examples, and the docs explain how jobs work.

Verdict

PixVerse V6 Max is a focused upgrade rather than a new generation: the same flexible 1 to 15 second format, now tuned for prompt following, realistic materials and lighting, legible products and text, and above all dialogue that gives each character the right voice and lips that match. If those claims hold up in your tests, it fixes the single biggest complaint about V6. Until then, treat it as the PixVerse model to try first for talking scenes, music-led clips and product shots from a single image, and keep the rest of the V6 family for references, transitions and multi-clip cuts.