PopNow Python SDK: AI Video, Image and Music Generation API

popnow is a typed Python client for the PopNow API that turns text, images and audio into AI videos, images, music and speech, using the same models, prices and credits as popnow.ai.

PopNow is a browser-based AI studio for videos, images and music: image-to-video, text-to-video, AI images, music and speech, one-tap trending video effects and image styles. This SDK wraps its developer API (https://api.popnow.ai/api/v1) so you can call Kling, Seedance, Wan, Vidu, MiniMax, Nano Banana, GPT Image, Flux, Suno and more from Python. Credit packs, no subscription; failed runs are refunded. Make it pop.

• Sync and async clients (PopNow, AsyncPopNow) built on httpx, the only dependency.
• Typed results (py.typed) with snake_case attributes, plus the raw JSON on every object.
• One-line helpers: client.images.generate(), client.videos.generate(), client.audio.speech().
• Safe retries: automatic Idempotency-Key, back-off on 429 / 500 / 503 honouring Retry-After.
• Local file uploads, asset library, task history, usage, balance and webhook signature verification.

Quickstart

1. Install (Python 3.9+):

pip install "git+https://github.com/popnow-ai/popnow-python"

A PyPI release is coming; until then, install from GitHub as above (pin a commit or tag with @<ref> for reproducible builds).

2. Create an API key in the API console and export it:

export POPNOW_API_KEY="sk-popnow-..."

3. Generate:

from popnow import PopNow

client = PopNow()  # reads POPNOW_API_KEY

task = client.images.generate(
    model="alibaba/z-image-turbo/text-to-image",
    prompt="A small red fox curled up in a teacup, soft morning light",
)
print(task.state, task.credits, task.urls[0])
client.files.download(task.assets[0], "fox.png")

That image costs a couple of credits (1 credit = $0.01). Buy credits on the pricing page; API calls and the website share one wallet.

What you can call

Supported models

Model ids look like <maker>/<model>/<task>. Some examples; the live list with prices is client.models.list() or popnow.ai/docs/models:

Every model has its own input_schema (client.models.get(id).input_schema). Unknown fields are rejected with a 400, so typos fail fast.

Usage

Image to video

import pathlib
from popnow import PopNow

client = PopNow()
task = client.videos.generate(
    model="alibaba/wan-3.0/image-to-video",
    image_url=pathlib.Path("photo.jpg"),  # a Path is uploaded for you; URLs work too
    prompt="Waves roll in slowly, the camera drifts toward the lighthouse",
    resolution="480p",
    duration_seconds=5,
    timeout=1800,
)
client.files.download(task.assets[0], "clip.mp4")

Music, speech and transcription

song = client.audio.music(model="suno/v6/text-to-audio", prompt="Upbeat lo-fi beat for a cooking video", instrumental=True)
voice = client.audio.speech(model="google/gemini-3.8-flash-tts/text-to-speech", text="Welcome back!", voice_id="Kore")
transcript = client.audio.transcribe(model="xai/grok-stt/speech-to-text", audio_url="https://example.com/interview.mp3").text

The helpers fill the shared field names (prompt, text, image_url, image_urls, video_url, audio_url, duration_seconds, resolution, aspect_ratio) and pass every other keyword straight into the model input. They wait for the result and raise TaskFailedError unless the task succeeds.

Low-level task API

from popnow import PopNow, new_idempotency_key

client = PopNow()
body = {
    "model": "alibaba/wan-3.0/image-to-video",
    "input": {"image_url": "https://media.popnow.ai/movie/lighthouse.webp", "resolution": "720p", "duration_seconds": 5},
}

quote = client.jobs.quote(**body)                 # nothing is created or held
created = client.jobs.create(
    **body,
    idempotency_key=new_idempotency_key(),        # reuse it on every retry of this work
    quote_id=quote.quote_id,
    expected_credits=quote.max_credits,           # QuoteMismatchError (40901) if the price changed
    callback_url="https://your-app.example.com/hooks/popnow",
)
task = client.jobs.wait(created, timeout=1800)    # polls with back-off: 2s, 3s, 4.5s ... up to 15s
if task.is_succeeded:
    print(task.urls)
else:
    print(task.state, task.error_code, task.error_message)

• jobs.create(..., wait=60) holds the connection up to 60 seconds; if the task finishes in time, created.task is the full record.
• jobs.create_and_wait(model=..., input=...) combines the server-side wait with polling.
• Task states: queued, running, then succeeded, failed or expired. Unknown states count as not final.
• output.assets[].url is a signed link valid for about 20 minutes. Store the task_id, not the link.

Uploads and the asset library

uploaded = client.files.upload("reference.png")                          # popnow://f/fil_...
asset = client.assets.save_upload(uploaded.file_id, name="Product shot")  # popnow://a/ast_...
for a in client.assets.iter(kind="image", sort="recent"):
    print(a.name, a.uri)
lora = client.files.import_lora("your-name/your-lora-repo", token="hf_...")  # for loras[].path

Uploads are temporary; saved assets stay until you delete them. Images up to 10 MiB, video and audio up to 90 MiB (600 seconds), LoRA .safetensors up to 2 GiB.

Async

import asyncio
from popnow import AsyncPopNow

async def main() -> None:
    async with AsyncPopNow() as client:
        balance = await client.account.balance()
        task = await client.images.generate(model="alibaba/z-image-turbo/text-to-image", prompt="A paper boat")
        print(balance.available_credits, task.urls)

asyncio.run(main())

Webhooks

import os
from popnow import webhooks, WebhookVerificationError

def handle(raw_body: bytes, headers: dict) -> int:
    try:
        event = webhooks.verify_task_callback(raw_body, headers, os.environ["POPNOW_WEBHOOK_SECRET"])
    except WebhookVerificationError:
        return 401
    task = webhooks.task_from_event(event)  # event["request_id"] is the stable delivery id
    print(task.task_id, task.state)
    return 200

Task callbacks are signed as base64(HMAC-SHA256(secret, "<taskId>.<timestamp>.<sha256_hex(raw_body)>")); timestamps more than five minutes away are rejected. Account endpoints use the Standard Webhooks format: webhooks.verify_endpoint_event(raw_body, headers, "whsec_..."). Always verify the raw request bytes.

Configuration

client = PopNow(
    api_key="sk-popnow-...",           # default: $POPNOW_API_KEY
    base_url="https://api.popnow.ai",  # default: $POPNOW_BASE_URL or this
    timeout=60.0,                      # seconds per request, or an httpx.Timeout
    max_retries=2,                     # network errors and codes 429, 500, 503, 50301
    language="de",                     # Accept-Language for error messages
    http_client=None,                  # bring your own httpx.Client (proxies, mounts)
)

Keep keys on your server; never ship them in a browser or mobile app.

Error handling

Every error response is {"code", "msg", "request_id"}. The SDK raises a typed exception chosen by the business code:

All API errors carry code, status_code, message, request_id and, for some asset and content errors, reason (data.reason).

from popnow import APIError, InsufficientCreditsError, RateLimitError

try:
    client.jobs.create(model="kling/3.0/text-to-video", input={"prompt": "A koi pond in the rain"})
except InsufficientCreditsError:
    print("Top up at https://popnow.ai/pricing")
except RateLimitError as err:
    print("retry in", err.retry_after, "seconds")
except APIError as err:
    print(err.code, err.message, err.request_id, err.reason)

Branch on code (and task.error_code), never on message text. Quote request_id when contacting support.

Examples

See examples/: generate_image.py, image_to_video.py, text_to_speech.py, quote_and_create.py, async_batch.py, save_to_library.py, webhook_receiver.py.

FAQ

How do I generate a video from an image with an API in Python?

Call client.videos.generate(model="alibaba/wan-3.0/image-to-video", image_url=..., prompt=...). Pass a public HTTPS URL, a popnow:// URI, or a pathlib.Path to a local photo, which the SDK uploads first. The call waits for the clip and returns the task with a signed download URL.

How much does the PopNow API cost?

The API uses the same credits and prices as popnow.ai: 1 credit = $0.01, bought in credit packs with no subscription. Each model lists its price in client.models.get(id).credits, and client.jobs.quote() returns the exact amount for one request. Failed and expired tasks release their hold, so they cost nothing.

How do I retry an AI generation request without being charged twice?

Send the same Idempotency-Key with every retry of one piece of work. The SDK generates one per call and reuses it across its own automatic retries; pass idempotency_key= yourself (for example order-8814-render-1) to stay safe across process restarts. Keys are remembered for 24 hours.

Do I have to poll for results?

No. jobs.wait() polls with back-off for you, jobs.create(wait=60) returns quick results in the same call, and in production you can pass callback_url and verify the webhook with popnow.webhooks.verify_task_callback().

How do I upload a local file to use as a reference image?

client.files.upload("photo.jpg") asks for an upload ticket, PUTs the bytes and commits them, then returns a uri (popnow://f/fil_...) for image_url or image_urls. To keep a file for reuse, save it with client.assets.save_upload().

Which AI video and image models can I use from Python?

Kling, Seedance, Wan, Vidu, MiniMax, Nano Banana, GPT Image, Flux, Seedream, Suno and more. Call client.models.list(modality="video") for the live list with availability and prices.

Is there an async Python client for AI video generation?

Yes. AsyncPopNow has the same resources and helpers as PopNow, with await. Combine it with asyncio.gather to run many tasks at once within the rate limits.

Can I use PopNow from an AI assistant instead of code?

Yes. PopNow runs a remote MCP server at https://api.popnow.ai/mcp with tools for images, video, music, speech and effects. See popnow.ai/mcp.

Links

• Website: popnow.ai
• Docs: popnow.ai/docs
• API reference (OpenAPI): popnow.ai/docs/openapi.yaml
• Pricing: popnow.ai/pricing
• MCP server: popnow.ai/mcp
• llms.txt: popnow.ai/llms.txt
• Other PopNow repos: popnow-mcp, popnow-skills, popnow-js, popnow-api, popnow-blender

Contributing

Issues and pull requests are welcome; see CONTRIBUTING.md. To report a vulnerability, see SECURITY.md.

License

MIT © PopNow