One API. Zero GPU Hassle. Pay per Generation, Not per Second.
The html-video project by nexu-io crossed 3,400 GitHub stars in just 26 days — a clear signal that developers are desperate for programmatic video. Existing solutions like RunwayML or HeyGen charge per second, lock you into proprietary formats, and offer no API designed for agents. They were built for human editors, not for code.
Worse, every AI agent framework — from AutoGPT to CrewAI — lacks a native video output. Agents can write text, generate images, even produce code, but they cannot create a simple product demo or social video without expensive, brittle integrations. This gap is absurd: video is the most engaging medium, yet it remains the hardest to generate programmatically.
Now is the perfect moment. The open-source html-video engine provides a free, local, HTML-to-MP4 pipeline with 21 templates and AI music. By wrapping it in a dead-simple POST /generate-video endpoint, we remove every barrier. No GPU, no CLI, no rendering farm — just JSON in, MP4 out. The window to own this “agent video API” category is open, and it will close fast.
Pick from 21 ready-made templates — product launch, social card, explainer, and more. Send a simple JSON payload with your headline, subheadline, and call-to-action text. No HTML or CSS required.
Your AI agent makes a single HTTP request to our endpoint. Include your API key and the JSON data. The engine renders the video using the open-source html-video library — locally, fast, and without any per-second licensing fees.
Within seconds, you receive a direct URL to the generated MP4 file. Stream it, download it, or embed it — no storage hassle. Your agent can now produce video content as easily as it writes text.
Powered by the open-source html-video engine, you pay only for the server resources you use — no per-second or per-video licensing. This makes high-volume agent video generation economically viable for the first time.
Product demos, social media cards, explainer videos, and more — all pre-built and tested. Just pass JSON with your content. No design skills, no HTML editing, no CSS tweaks needed to get a professional-looking video.
Simple POST /generate-video endpoint with JSON input and MP4 output. Works with any LLM, agent framework, or automation tool. No GPU, no CLI, no rendering farm — just a clean API your agent can call.