Fliz Logo
Sign upSign in

By the Fliz team · Updated November 12, 2026

AI video generation API: create complete narrated videos from your code

The Fliz API lets developers generate complete videos programmatically. Send a text, an idea or a URL, and the API returns a finished MP4 with an AI-written script, one generated image per scene, a voiceover in one of 32 languages, synced captions and music. It is a REST API documented with OpenAPI, included in every plan at no extra cost, with a hosted MCP server for AI agents and official nodes for n8n, Make and Zapier.

Create a video
Made with Fliz from the URL of a news article.

How the video generation API works

  1. 1

    Get an API key

    Create a key at app.fliz.ai/api-keys and send it as a Bearer token. The OpenAPI specification at app.fliz.ai/api/openapi describes every endpoint and field.

  2. 2

    Create a video and get its video_id

    Call create video with your text, or create video from URL with a page address, plus the generation options listed in the OpenAPI specification. The response returns a video_id right away.

  3. 3

    Poll the status or receive a webhook

    Check the status about every 60 seconds, or pass a webhook_url with the request. When the video is ready, get video returns its details and the MP4 to download.

A video API that returns finished videos, not clips

Most AI video generation APIs expose a clip model: you send a prompt and receive a few seconds of footage. That is useful for a shot, but to publish a video you still have to write a script, generate several clips that match, add a voice, captions and music, and assemble everything with a rendering pipeline of your own.

Fliz works one level higher. One API call produces a whole narrated video, usually a minute or more: the script, the scenes, the visuals in one consistent style, the voiceover, the captions and the music. If your product needs photorealistic cinematic footage, a clip model API is the better tool. If it needs explainer videos, social videos or product videos generated at scale, Fliz removes the whole assembly step.

Clip model APIs vs a video generation API

Developers comparing AI video generation APIs in 2026 usually look at clip model APIs first. Google Veo 3.1 is available through the Gemini API and Vertex AI and generates clips of 4 to 8 seconds with native audio. Kling 3.0 generates clips of 3 to 15 seconds with audio and lip sync, with credits charged per second of video. Runway Gen-4.5 generates video from a text prompt, a reference image or an existing video. The Sora API, from OpenAI, closed on September 24, 2026.

These models are text-to-video and image-to-video APIs: the output is a clip, and the price is per second or per generation. The Fliz API sits one level above them. The input is a text, an idea or a page, the output is a complete narrated video, and the cost is per video in credits. If your app needs a single cinematic shot, call a clip model. If it needs a finished video for each article, product or topic, Fliz saves you from building the script, voice, caption and rendering layers yourself.

What you can do with the API

The REST API and the MCP server expose the same set of operations:

  • Create video: from a text, an idea, an article or an ad brief, with your generation options.
  • Create video from URL: Fliz reads the page and writes the script from it.
  • Analyze URL: preview what Fliz extracts from a page, without spending any credit.
  • Get status and get video: follow the progress and fetch the result.
  • List videos: browse the videos of your account.
  • List voices and list musics: discover the voices and music tracks you can pass in a request.
  • Translate video and duplicate video: produce the same video in another language or as a new variant.
  • Credits balance: check how many credits remain before you launch a batch.

The asynchronous flow: create, poll or webhook

Video generation takes 2 to 20 minutes depending on the length and the options, so the API is asynchronous. A create request returns quickly with a video_id. From there, you have two ways to know when the video is ready.

Polling: call get status with the video_id about every 60 seconds until the video is complete, then call get video. This is the simplest option for scripts, cron jobs and batch processing.

Webhook: pass a webhook_url in the create request. Fliz calls that URL once when the video is ready. The call is made a single time with no retry, so if your endpoint may be down, keep a polling fallback for videos that have not reported back after a while.

Generation options you control

The create request carries the generation options, so each video in a batch can be different. The OpenAPI specification gives the exact field names; these are the choices Fliz offers:

  • Script style: 16 styles, including tutorials, product presentations, news, history, science and children’s stories.
  • Language and voice: 32 languages, a male or female voice, or a custom voice id.
  • Visual style: one of 63 styles, from realistic to illustrated, with optional motion on the hook or the whole video.
  • Format: 9:16, 1:1 or 16:9, MP4 in 1080p on paid plans.
  • Captions: 4 animated styles, center or bottom, 10 fonts, any color, optional uppercase.
  • Music: a track from the library or your own, with adjustable volume, and your logo as a watermark.

Rate limits and pricing

The REST API accepts 300 requests per minute, and the MCP server 60 requests per minute. Because a video takes minutes to render, the limit that matters for most integrations is your credit balance, not the request rate: check it with the credits balance endpoint before you queue a large batch.

There is no separate API price. The API is included in every plan, and videos created through it use the same credits as videos created in the app: a simple video uses 95 credits. A new account gets one free trial video, in 720p with the Fliz watermark, which is enough to test the output before you integrate.

MCP server for Claude, ChatGPT, Claude Code and Cursor

Fliz hosts an MCP server at app.fliz.ai/mcp, over Streamable HTTP, with OAuth 2.1 or an API key sent as a Bearer token. Add it as a custom connector in Claude or ChatGPT, or in Cursor, and the assistant can create videos, check their status, list voices and music and translate a video for you in plain language.

In Claude Code, one command adds it: claude mcp add --transport http fliz https://app.fliz.ai/mcp. It is a fast way to prototype a workflow with an AI agent before you write the integration code.

No-code: n8n, Make, Zapier and WordPress

Not every integration needs code. Fliz has an official n8n community node, published on npm as n8n-nodes-fliz, an official Make module and an official Zapier app. A WordPress plugin creates a video when a post is published.

Typical scenarios: a new row in Google Sheets generates a video that is then published on TikTok; a new RSS item becomes a video posted on Instagram; every day at 3pm, a children’s story is generated and published on YouTube. Fliz creates the video, and the automation tool handles the publishing step, since Fliz does not post to social networks itself.

How to choose an AI video generation API

Before you build on any video generation API, compare these points on a real request, not on a demo page:

  • Output: a clip of a few seconds, or a complete video ready to publish?
  • Input: text prompts, a reference image for image-to-video, or your own content such as a URL?
  • Pricing model: per second, per generation or per video, and what one finished video costs once retries are counted.
  • Latency: how long a generation takes, and whether the API is asynchronous with polling or a webhook.
  • Resolution and formats: 720p or 1080p, and vertical, square or horizontal output.
  • Audio: native sound, a voiceover, or nothing at all.
  • Rate limits and access: requests per minute, API key or OAuth, and the plans that include API access.

Use cases for developers and product teams

Media sites turn each new article into a short video for social media. E-commerce platforms generate a product video from a product page or from 3 to 10 product photos, for every item in a catalog. SaaS products add a “turn this into a video” feature for their users. Agencies run the same brief in several languages with translate video. Content teams feed a queue of topics into the API and review the scripts before publishing.

Questions

Is the Fliz API included in every plan?

Yes. The REST API is available on every plan at no extra cost. Videos created through it use the same credits as in the app.

Where is the API documentation?

The OpenAPI specification is at app.fliz.ai/api/openapi, and API keys are created at app.fliz.ai/api-keys.

Is the API synchronous?

No. A create request returns a video_id; then you poll the status about every 60 seconds or pass a webhook_url, which is called once when the video is ready, with no retry.

What are the rate limits?

300 requests per minute for the REST API and 60 requests per minute for the MCP server.

Can the API publish to YouTube or TikTok?

No. The API creates the video. Publishing goes through your own code, your scheduler, or an automation in n8n, Make or Zapier.

Can I use my own voice?

Yes. Through the API you can pass a custom voice id, or pick one of the voices returned by list voices.

Does analyzing a URL cost credits?

No. Analyze URL previews what Fliz extracts from a page without spending any credit.

Generate your first video through the API

Create an account, get an API key and send your first request: the trial video lets you check the output before you build on it.

AI Video Generation API: Narrated Videos via REST | Fliz