How to Use PixVerse

PixVerse V6 turns a sentence into a clip of up to 15 seconds at 1080p. A first run is four decisions: which model, which input, three output switches, then refinement. This guide uses the vendor’s numbers and marks what PixVerse has not published.

Start with the right model: PixVerse V6, C1 or R1

Three video models carry the PixVerse name: C1 and V6 are generation models, selected through the model parameter, and R1 is not in that selector.

Video extension is the one hard difference: unsupported on C1, supported on V6. Everything else a first render needs sits on both, at the same per-second price. C1 adds narrative instead: storyboards built from a prompt, and static storyboard layouts converted into continuous sequences.

R1 is not a delivery model. PixVerse’s test: use R1 when the output should behave like a live world, a standard video model when the output should be a finished MP4 — if the deliverable is a file, choose V6 or C1. R1 starts from world.pixverse.ai; the same question for another family is in the Seedance hub.

The three switches that decide what a clip costs

PixVerse charges per second, and three settings fix the rate: quality (360p, 540p, 720p or 1080p), duration (1–15 seconds) and generate_audio_switch, off by default.

Those are credits per second. The spread on one unchanged prompt is 4.6×: 5 at 360p without audio, 23 at 1080p with audio. Two multipliers sit on top: Fusion’s video_references input doubles the V6 rate (360p 10/14, 540p 14/18, 720p 18/24, 1080p 36/46), and C1 costs one credit per second more than V6 at every tier.

The only currency conversion published is $1 = 5 videos at v6, 720p, 5s, no audio, with the Starter pack. Plan prices, monthly credit grants and new-account credits are not published.

Your first clip, step by step

The V6 model page reduces a first run to three steps.

1. Choose V6 and the input: text-to-video for a new concept, image-to-video for an existing asset.
2. Set the output format before generating: duration, resolution and aspect ratio.
3. Generate, then refine, reviewing character consistency, motion stability, lens behaviour and sound timing.

Prompts work best between 25 and 200 words, and image-to-video asks whether the camera pushes in, pans, tilts, tracks or circles. PixVerse quotes renders at an average of 10 seconds, as fast as 5, without saying which model or resolution that was measured on.

The PixVerse hub gathers the model line; the code path — key, endpoints, polling, and credits separate from a web account — is in the PixVerse API guide.

A worked example: one prompt, two settings

The credit rate moves 4.6× between the two settings.

Case one: the vendor’s own prompt, at the cheap end

This prompt is published on the V6 model page, and already carries the camera instruction the image-to-video guidance asks for.

Prompt: A matte black smart speaker sits on a walnut desk at sunrise. The camera slowly pushes in from the left. Warm light reflects on the metal rim. A soft startup chime plays as the LED ring turns blue.
Model: v6
quality: 360p
duration: 5
aspect_ratio: 16:9
generate_audio_switch: false

At 360p without audio that is 5 credits per second: five seconds, 25 credits.

Case two: the same prompt, at the expensive end

Prompt: (unchanged)
Model: v6
quality: 1080p
duration: 15
aspect_ratio: 21:9
generate_audio_switch: true
seed: 0

Same words, 23 credits per second, 15 seconds instead of 5. Add a video_references input and the 1080p with-audio rate doubles again, to 46 credits per second. Only resolution, duration and the audio switch moved.

Parameters and hard limits

Prompt length is where PixVerse contradicts itself, and both figures are official: the model pages document 5000 characters, the error-code page describes codes 400018 and 400019 as exceeding a 2048-character limit. The matrix caps both models at 15 seconds and 1080p, so 4K is not on offer, and it publishes nothing on concurrency, failure rates or regional availability.

What you can do after the first render

Style repainting, subject replacement, motion mimicry, prompt-driven edits, synchronised sound effects, lip sync and template-based image generation are documented as standalone capabilities, each with its own endpoint, inputs, pricing and limits. Upscale arrived on 16 June 2026.

Extension splits the two generation models: V6 continues an existing clip, C1 cannot, so a project that will outgrow its first render belongs on V6.

Output rights on the web path

The terms behind pixverse.ai are operated by AIVORA PTE. LTD. and were updated on 7 May 2026. PixVerse does not claim ownership of the outputs; within non-commercial use you keep ownership of your inputs and all rights in the outputs, and use is limited to non-commercial purposes unless you obtain a separate authorisation or commercial-use licence. Platform creative materials are restricted to in-platform creative use, and the service is for people aged 16 and over. The API runs under a different operator, with its own terms.

Frequently asked questions

Is PixVerse free to try? No plan prices and no new-account credit figure are published, so plans cannot be compared from official numbers. The rate card is public: 360p without audio is the cheapest V6 tier, at 5 credits per second.

Which model should I use first? V6 for a first clip, C1 when the project is built from storyboards, R1 only when the output should behave like a live world.

Why is my prompt rejected as too long? Two official limits disagree: 5000 characters on the model pages, 2048 on the error-code page. Stay under the lower figure.

Can I extend a clip made with C1? No. Video extension is unsupported on C1 and supported on V6.

Decide the rate before you write the prompt.

Published

Sep 14, 2026

Information verified

Sep 14, 2026

By

AI Tool Blog Editorial Team