GPT-Video for developers

GPT-Video turns a long video into short vertical clips with captions and face framing. The GPT-Video API is a video editing API for your own code. Your program uploads a video, GPT-Video transcribes it and proposes the moments that work as clips, and your program exports the clips it keeps as MP4 files.

AI assistants such as Claude and ChatGPT can use GPT-Video through the MCP server. The API and the MCP server both act on your GPT-Video account.

API bonus: 50 credits. No card needed. After the bonus, the API needs a Creator or Studio plan. API pricing explains what each call uses.

How to call it

The base URL is https://api.gpt-video.com/v1. Every request sends an API key in the Authorization header as Bearer gvk_.... You create keys on the Developers page of the GPT-Video app. Authentication has the details.

The OpenAPI 3.1 document is at https://api.gpt-video.com/v1/openapi.json. There is no official SDK yet. The OpenAPI document works with code generators.

Every endpoint is in beta. It is supported and can gain new fields. If a change would break your code, we tell you before it takes effect and list it in the changelog.

How the pieces fit together

An asset is a video file you uploaded. A project points at one asset and holds the settings. Analyzing a project transcribes the video and proposes clips. Accepting a clip makes a composition, which is the vertical video with its captions and framing. Exporting a composition renders an MP4 file, which you download from the export. Analyses and exports run as jobs, and you wait for a job to end before you read its result.

Objects describes each of these and the fields your code reads.

Start here

Quickstart takes you from an API key to a downloaded clip.
Use GPT-Video from an AI agent has a prompt to copy and the rules an agent should follow.
Authentication covers keys, scopes and the errors a key can cause.
API pricing lists what each call uses and how the API bonus works.

Guides

Concepts

Errors lists every error code and what your code should do next.
Rate limits and daily limits gives the limits per key and per account.
Pagination and idempotency explains cursors and the Idempotency-Key header.
Jobs shows how to wait for an analysis or an export.
Webhooks lists the events GPT-Video can send to your server.

API reference

The API reference has one page for each endpoint. Each page shows the parameters, a request in curl, TypeScript and Python, the response, the errors and the cost.

For AI tools

The API manual is one Markdown file with the conventions, the error table and a line for each endpoint. An AI agent should read it first.
skill.md is a skill file an agent can load before it calls the API.
Every page of these docs also exists as Markdown at the same address with .md at the end.

The MCP server page explains the two editions and how to connect Claude, Cursor and other assistants to your GPT-Video account.