Sign up and get free credits

Feature

The Vidonto API and webhooks

Make videos from links, text, scripts or documents in code, review them before rendering, manage brands, edit and export recordings, with signed webhooks.

Updated

Most of what Quick Render does, you can do from your own code. The Vidonto API turns YouTube links, web pages, text and your own scripts into finished videos, with the same settings, the same credits and the same results as the app. Use it to make a video for every new blog post, every episode of a podcast, or every product in a catalogue.

The API keys section of the Vidonto Account page, with a form to create a named key and a list of keys with their creation date
Create and revoke API keys on the Account page.

What you can do

  • Start a run from a link (YouTube or web page), from pasted text, from your own script, or from a PDF, Word file or PowerPoint deck, with options for length, video size, on-screen text, caption length, AI images, opening and ending, instructions, crediting the source, the voice, the video language and your brand look.
  • Make videos in other languages. Send videoLanguage (for example "es" or "de") and the script, on-screen text, voice-over and captions are in that language. The options endpoint lists the 26 languages; without it, runs use your saved default, or English. The price is the same in every language.
  • Use your own script. Send script instead of a link or text, and your words are narrated exactly as written, up to 20,000 characters. There’s no AI script step and no script charge, and the storyboard splits the script into scenes without rewording it. Options that only steer the AI script — opening, ending, sign-off, crediting the source, the script model and the target length — are refused with the error code own_script_option. Instructions still steer the storyboard.
  • Combine several sources into one video. Send sources, a list of 2 to 5 links, texts or document links (documentUrl, a PDF, Word file or deck) in the order you want them, instead of a single link or text, and an optional writingMethod: new_piece (the default) for an original video that uses the sources as research, or summary to retell what they say. Text and document sources can carry their own title and author or site for the source list. Documents are read, free, before the run starts and never stored. The links are read when the run starts, each YouTube video’s captions being a paid read. Estimate it with sourceKind=multi, sourceCount and youtubeSources. See several sources, one video.
  • Start from just a topic. Send topic (up to 200 characters) and an optional angle (up to 300) instead of any source, with an optional writingMethod. Add videoType (how_to, feature_tour, comparison or explainer) to shape the search, the script’s structure and the storyboard’s scenes; it costs nothing extra, and the run’s options.videoType shows it (null when none was picked). GET /options lists the types under topic.videoTypes. The run’s first step searches the web and picks 2 to 5 readable articles and YouTube videos itself, then makes the video. The picked sources are listed in the run’s topic.sources. Estimate it with sourceKind=topic: the estimate covers the search and the whole video. If fewer than 2 readable sources turn up, the run fails with a retryable error and nothing is charged for the search; a retry never searches twice. To make several topic videos at once, send topics (up to 20) to the batch endpoint instead; they share one angle, writingMethod and videoType.
  • Make a highlights video of a recording. Send recordingId (from the recordings list) and an optional highlightsStyle, narrated (the default) or cards, instead of a link or text. The recording must be kept in the cloud and transcribed, or the request is refused with highlights_recording before anything is charged. The AI picks its strongest moments as sound bites, which play with the speaker’s own sound, joined by short spoken bridges or text cards; the voice is charged only for the bridges, and cards have no voice charge. Opening, ending, sign-off, crediting the source and the AI presenter are refused with highlights_option. In a reviewed run, a sound bite’s words can’t be edited (sound_bite_words); edit the bridges around it. Estimate it with sourceKind=recording and highlightsStyle. See highlights videos.
  • Start from a document. Send documentUrl, a public link to a PDF, Word (.docx) or PowerPoint (.pptx) file up to 50 MB, and the run starts from its text, read the same way as in the app. The file is read and never stored. For a file on your computer, send it to the document reader endpoint first and start the run with the text it returns. Files that can’t be read, such as scanned pages without text or password-protected files, are refused before anything is charged.
  • Review before it’s made. Send review: true and the run pauses twice: once its script is written and again once its storyboard is ready. Read what’s waiting, change the title, script or individual frames (their words, on-screen text, image prompt and where the presenter stands), then approve, and the run carries on. The approved storyboard’s words are exactly what the voice-over says. A topic run pauses once more before that, after its search: its picks and the search’s other readable suggestions are listed, and you can drop a pick or swap in another suggestion (keeping 2 to 5) before approving the sources, the same step as Check the sources first in Quick Render. Reviewing costs nothing extra, and a run from your own script only pauses for the storyboard. Subscribe a webhook to run.review_waiting and you’re told the moment each step is waiting, instead of checking.
  • Add an AI presenter to a run or a batch: a realistic presenter who speaks the narration on camera. Send an avatar id and a layout — corner, bookends (opening and ending), full, or director, which places the presenter scene by scene like Let the director decide in the app. The corner presenter’s side (left or right) and size (small, standard or large) can be set too, and for corner and bookends a stage: beside the scenes or in the scene. The options endpoint lists the presenters, layouts, price per minute and the longest narration allowed (10 minutes). See AI presenter.
  • Use your brand. Send look: "brand" to use a brand’s colours and fonts saved in Settings instead of the built-in look, and showLogo: true to put its logo in the corner. Runs use your default brand unless you send brandId to pick another one, such as a client’s. The options endpoint lists your brands with their IDs and whether each has colours, a logo and fonts saved. A brand that isn’t yours or was deleted, or one missing the colours or logo asked for, is refused before anything is charged. Each run keeps the brand it started with, even if you change it in Settings later, and the look costs nothing extra. See brand kit.
  • Keep one look. Leave look out and the AI picks a built-in look for each video, so it can change. Send a built-in look such as "midnight" or "paper" to use it every time, or save a Summary video look in Settings and runs that leave look out use it; look: "auto" lets the AI pick for one run. The look is kept on retries and costs nothing extra.
  • Manage brands, saved intros and saved styles — the same ones as in Settings and the Video Editor. Add, change and delete brands, upload or remove their logos, and save, rename or delete intros and styles, with the same rules and limits as the app.
  • Start a batch of up to 100 links, or up to 20 topics, with the same options. Every link or topic is checked before anything starts, and credits for the whole batch (for topics, each search plus its video) are held at once, or nothing starts.
  • Get an estimate in credits before you start, and read your balance and credit history.
  • Track runs — list them, get the status and current step of one, cancel it or retry it after a failure.
  • Download the video when the run has succeeded.
  • Receive webhooks when a run succeeds, fails or is cancelled, when a recording is transcribed or fails, or when an export finishes or fails, instead of polling.
  • Add recordings to the Video Editor. Upload a video file (MP4, MOV, WebM or MKV, up to 16 GiB and 6 hours) in parts that can be sent in any order and resumed, or send a YouTube, TikTok, X or Facebook video link (confirming you own the video or have permission to use it) or a direct download link and Vidonto fetches it. The video is kept in the cloud and transcribed, so it opens in the Video Editor on any device, ready to edit by text. An estimate endpoint shows the credits first.
  • Read your recordings — list and search them, filter by tag, check one’s status, length and chapters, and read its transcript with your fixes to misheard words applied, with or without each word’s timing. Set tags and delete recordings too. Recordings made in the app are included.
  • Edit recordings like the Video Editor. Run its AI tools — find mistakes, Direct my video, suggested scenes, chapters and short clips — with an estimate first, then accept all of their suggestions, some by ID, or dismiss them. Cut or restore words or exact times, fix misheard words, and place camera views, zooms, scenes, text and logos or a saved intro. Change a recording’s AI model, background, AI avatar, camera shape and polished video options, and switch the cleaned-up sound on or off and set its strength.
  • Finish recordings. Dub a recording into up to five languages at a time; render the edited video in the cloud at the original size, 1080p or 720p; make a polished video with the options saved for it; or export one short clip, with the original sound or a finished dub. Each has an estimate first, is charged once when it finishes and is downloaded from the API. Also get SRT or WebVTT caption files for the edited video or a dub, write the description and social posts or a blog post or how-to article, read the chapters, and make, list or end share links.

How it works

  1. Create an API key in the app under Account → API keys. Give it a name you’ll recognise. You can have up to 10 active keys and revoke any of them at any time.
  2. Send requests to https://www.vidonto.com/api/v1 with the key as a bearer token in the Authorization header.
  3. Start a run, then either poll the run until it finishes or register a webhook to be told.
  4. Download the MP4 from the run’s video address.

Every request that starts paid work can carry an idempotency key. If your request is retried because of a network error, the same run is returned instead of starting — and charging for — a second one.

The Quick Render runs list showing a finished run with its four steps, which matches the steps an API run reports
API runs go through the same steps as Quick Render runs.

Recordings

A YouTube, TikTok, X or Facebook link is read with the same checks as From a link in the app: one public video, not a channel or playlist. Send rightsConfirmed: true to confirm you own the video or have permission to use it, or the link is refused with rights_not_confirmed. Instagram and Vimeo links can’t be read (site_not_supported), and private, age-restricted, members-only and live videos are refused too. The estimate reads a page’s length and size, and a refused link holds and charges nothing.

Adding a recording works like keeping one in the cloud from the app. The storage price for the original and an editing copy is held when it starts and charged on the actual size once the recording is transcribed; it covers the first 30 days and then renews every 30 days. Transcription and the automatic mistake check are charged on the actual length, as in the app. If the file can’t be read, has no sound or can’t be transcribed, storage isn’t charged and the files are deleted.

Editing uses the Video Editor’s own rules, and changes show up in the app straight away. Each change sends the revision it was based on: if the recording was changed since, in the app or by another call, nothing is saved and you read the edits again. A list of edits is saved all together or not at all, and Accept all works exactly as it does in the editor — nothing already placed is replaced, and suggestions that don’t fit are left waiting. Only the AI tools spend credits: the estimate is held and the measured cost charged when the tool finishes, and repeating a request with the same idempotency key never charges twice. A background effect is used in exports only after the recording has been opened in the Video Editor, which checks that someone is on camera.

Recordings made in the app are listed too. When the original video is only on the computer it was recorded on, the recording says so: you can read its details and transcript, but anything that needs the video itself is done in the app on that computer.

Webhooks

Register an HTTPS address and choose the events you want: run.succeeded, run.failed, run.cancelled, run.review_waiting, recording.ready, recording.failed, export.ready and export.failed. Each delivery is signed with a secret only you and Vidonto know, so you can check that it really came from Vidonto. You can send a test event, see recent deliveries, resend one and rotate the secret.

No code? Use an AI assistant

If you’d rather ask for videos in plain words, connect Vidonto to Claude, ChatGPT, Cursor or another assistant that supports MCP connectors. You sign in and approve it once instead of handling a key, and it checks the credits with you before starting anything. See use Vidonto from an AI assistant.

Full documentation

The developer docs have the full reference: a quickstart in curl, JavaScript and Python, every endpoint and option, error codes, rate limits, idempotency rules and how to verify webhook signatures.

Limits worth knowing

  • Runs without review are approved automatically, like Quick Render. With review, frames’ on-screen points follow the narration and can’t be edited one by one; for that, use the Studio’s storyboard editor.
  • Exports render from the video kept in the cloud, and dubbing needs the recording’s sound prepared; recordings added through the API are ready for both, while ones made in the app need Keep in the cloud and opening once in the Video Editor. Cleaning up the sound itself is done in the app; the API switches it on and sets the strength.
  • Custom animated scenes are made by the storyboard; with review you can switch a frame to or from one, but not write your own.
  • Requests are rate-limited. Use webhooks rather than polling often. The docs list the limits.
  • Keep keys secret. Anyone with a key can spend your credits. Revoke a key straight away if it leaks.

Credits

API runs and recordings use the same balance and the same prices as the app. A run or recording that can’t be covered by your balance is refused before it starts. See credits explained.

Ready to build? Follow make videos in bulk with the API.