What Pixverse v6 is

Pixverse v6 runs on fal.ai as fal-ai/pixverse/v6/text-to-video and fal-ai/pixverse/v6/image-to-video. It's one of two Pixverse models on the platform (the other is C1, which is the cinematic-focused sibling). v6's identity: broad aesthetic range through named style presets, an internal prompt optimizer called thinking_type, and pricing that makes batch work economically trivial.

What it does well

Style presets are the headline. You can pass a named style value and the model applies that aesthetic without needing to encode it in the prompt text. That keeps prompts focused on subject and action. The thinking_type parameter runs an optional internal optimization pass over your prompt before generation - enabled for complex prompts, disabled for speed, auto for the model's call.

Resolution ladder goes 360p, 540p, 720p, 1080p - four tiers at the same per-second cost. Eight aspect ratios including 21:9, 2:3, and 3:2. Duration is an integer from 1 to 15 seconds, defaulting to 5. Audio is off by default but can be toggled on via generate_audio_switch (yes, the trailing _switch is real - Pixverse uses that naming convention). Negative prompts are supported; seed is supported.

What it can't do

No 4K, same as most mid-tier models - 1080p is the ceiling. Audio generation, while available, is less mature than Veo or Seedance. No end-frame or video-continuation parameters, so multi-shot sequences require external stitching.

Parameters that matter

The image-to-video endpoint adds image_url as required.

Pricing

$0.03-$0.12 per second (tiered by resolution and audio). That is still genuinely cheap for batch work. A 10-second 1080p clip with no audio costs $0.95 (10 x $0.095). A 15-second 1080p clip with audio caps at $1.80 (15 x $0.12). For batch workflows running hundreds of clips per day, Pixverse v6 lives in a different price tier than everything else on fal.ai.

Resolution is a cost lever as well as a quality lever. Drop to 360p or 540p for drafts and the per-second rate falls to $0.03-$0.04, so draft volume gets cheap fast.

Working example

1import { fal } from "@fal-ai/client";
2
3const result = await fal.subscribe("fal-ai/pixverse/v6/text-to-video", {
4 input: {
5 prompt: "A neon-soaked alley in a rainy megacity, a courier dashes past glowing storefronts",
6 style: "cyberpunk",
7 thinking_type: "enabled",
8 resolution: "1080p",
9 aspect_ratio: "16:9",
10 duration: 8,
11 generate_audio_switch: true,
12 negative_prompt: "blurry, pixelated, watermark",
13 seed: 42,
14 },
15});
16
17console.log(result.data.video.url);

When to pick Pixverse v6

Pick it for stylized output (anime, clay, cyberpunk, comic, 3d_animation presets), batch generation at scale, or any workflow where the per-generation cost needs to be near-zero. Skip it for 4K requirements, and skip it when audio is a hero requirement (Veo 3.1 or Seedance handle audio more reliably). For prototyping or high-volume creative ideation, the thinking_type: "auto" + style preset combo is a sensible default.