PixVerse API Pricing: Credits, Rate Card and What a Batch Costs

PixVerse publishes one page of API prices, and everything below comes from it. Where that page stops, this article stops too, because a rate nobody published is not a rate.

Which PixVerse price list you are paying

The subscription guide opens with warnings rather than instructions: API memberships are separate from PixVerse Web (app.pixverse.ai) Membership and API credits cannot be used with the PixVerse Web. Same login, different wallet. Plan changes are marked Not supporting yet.

Most numbers circulating online belong to the other wallet. Everything numeric here is API-ledger; the consumer credit page covers that side, the API guide covers auth and endpoints.

The per-second rate card, quoted

The first model table is headed Billing standards, with the unit column reading 1s. Generation bills per second of output, not per request.

V6 and C1 per second

The Fusion doubling

A Fusion request carrying video_references gets its own table at double the V6 rate: 360p 10/14, 540p 14/18, 720p 18/24, 1080p 36/46. C1 has no multiplier and sits one credit above V6 without audio.

Not everything bills per second. Sound effects are +2 Credits per 1 second. Restyle is 10 credits per second, Upscale 5, Lip sync 4 per second of external audio or per 15 bytes after UTF-8 encoding of TTS, Swap mask selection 2 per request. Multi-transition is tiered, 23 credits at one second to 270 at thirty. V5.5 and V5.6 are priced per clip, and on V4.5 down to V3.5 When motion_mode is set to fast, credit consumption is doubled (×2).

Membership plans and what a pack buys

Credits arrive two ways. A membership grants a monthly allowance; a pack is bought outright.

The membership video counts are the vendor’s, footnoted Calculated based on V3.5 Turbo / 540P 5s video. Every pack row lands on one cent per credit; the membership rows do not, so Business buys credits more cheaply than Essential. The headline names a Starter pack that appears in neither table.

Which wallet a credit sits in is readable from the API.

# GET /openapi/v2/account/balance returns both balances separately.
curl --location --request GET 'https://app-api.pixverse.ai/openapi/v2/account/balance' \
  --header "API-KEY: ${PIXVERSE_API_KEY}" \
  --header "Ai-trace-id: $(uuidgen)"
# credit_monthly resets every month; credit_package is what you bought outright.

That docs page shows ai-trace-id: 1 in its example, while the quick-start requires a unique UUID per request.

Three published anchors that do not agree

The pricing page states $1 = 5 videos (v6, 720p, 5s, no audio, with Starter pack). Per second that is four cents; at 9 credits per second and one cent per credit it is nine. Run the conversion rather than eyeballing it.

# What the headline anchor implies, against the published rate card.
seconds_per_dollar=$((5 * 5))   # 5 videos of 5 seconds
credits_per_second=9            # V6 720p, no audio
awk -v s="$seconds_per_dollar" -v c="$credits_per_second" 'BEGIN {
  printf "headline anchor: $%.3f per second, $%.2f per minute\n", 1 / s, 60 / s
  printf "rate card at $0.01/credit: $%.2f per minute\n", c * 0.01 * 60
}'

That prints $0.040 against $0.090 for the same second of the same model. The official homepage adds a third: its affordability chart, labelled Source: Artificial Analysis and dated as of 2026-04-02 0:00 PST, lists V6 at $4.80/min.

Every figure in the last column is our arithmetic on published numbers. Which row binds depends on how you buy, and the pack price is the only one that applies to a small test.

Estimating a batch from published numbers

The script below takes the published card and a retry count, because a first render is rarely the one you keep.

#!/usr/bin/env python3
"""Estimate a PixVerse API batch from the published per-second credit card."""

# credits per second: model -> quality -> (no audio, with audio)
RATES = {
    "v6": {"360p": (5, 7), "540p": (7, 9), "720p": (9, 12), "1080p": (18, 23)},
    "c1": {"360p": (6, 8), "540p": (8, 10), "720p": (10, 13), "1080p": (19, 24)},
}


def clip_credits(model, quality, seconds, audio=False, fusion=False):
    no_audio, with_audio = RATES[model][quality]
    rate = with_audio if audio else no_audio
    if fusion:  # video_references doubles the V6 Fusion rate
        rate *= 2
    return rate * seconds


def batch(model, quality, seconds, shots, attempts, audio=False, fusion=False,
          usd_per_credit=0.01):
    per_clip = clip_credits(model, quality, seconds, audio, fusion)
    total = per_clip * shots * attempts
    return per_clip, total, total * usd_per_credit


if __name__ == "__main__":
    per_clip, total, usd = batch(
        model="v6", quality="1080p", seconds=5, shots=12, attempts=3, audio=True
    )
    print(f"{per_clip} credits per clip")
    print(f"{total} credits for the batch")
    print(f"${usd:,.2f} at $0.01 per credit")

Reading the estimator output

The defaults price twelve 5-second 1080p V6 clips with audio, three attempts each. That prints 115 credits per clip, 4140 credits for the batch and $41.40 at $0.01 per credit. Set attempts to your own retry rate rather than to one, and set usd_per_credit to whatever you actually paid; the rates are the published ones.

Concurrency and your schedule

The rate-limit page publishes no request-per-minute figure. It publishes simultaneous generating tasks.

Backpressure instead of a rate limit

Concurrency decides wall-clock time, so estimate the wave count before promising a delivery date.

# Turn the published concurrency ceiling into a wall-clock estimate.
clips=120            # how many shots the project needs
seconds_per_clip=90  # observed average render time
concurrency=15       # Essential
waves=$(( (clips + concurrency - 1) / concurrency ))
minutes=$(( waves * seconds_per_clip / 60 ))
printf '%s clips at %s concurrent = %s waves, about %s minutes\n' \
  "$clips" "$concurrency" "$waves" "$minutes"

Exceeding the ceiling returns 500044, and the documented remedy is an upgrade or an email to api@pixverse.ai. Higher limits are part of the Enterprise Custom Plans route, which runs through api_business@pixverse.ai and is quoted as Custom.

What the pricing page does not publish

The content of the Starter pack, generation-time service levels, failure rates, regional price differences and any request-per-minute ceiling are not on any page we could open. Neither is a dollar figure for a finished project; that is arithmetic on the card above. The free plan page covers what is published about consumer limits, and the V6 and C1 comparison covers which model to price for.

Frequently asked questions

What does the PixVerse API cost per second? Nine credits for V6 at 720p without audio, up to 23 with audio at 1080p. C1 runs one credit higher than V6 without audio.

How many credits is a 5-second V6 1080p clip? 90 without audio, 115 with, on the published card.

Are API plans and web plans the same purchase? No. API memberships are separate from the web membership, and API credits cannot be used with the PixVerse Web.

Is there a request-per-minute limit? Not published. The limit is concurrent generating tasks, 3 on Free through 25 on Business; over it returns 500044.

What is the cheapest way to price a test? Use the pack price of $0.01 per credit and a 5-second 720p clip at 9 credits per second: about 45 cents.

Price from the page, not from a blog.

Published

Sep 17, 2026

Information verified

Sep 17, 2026

By

AI Tool Blog Editorial Team