# VidLab API

> One API key for the top image, video and audio models. Each call goes to the best available route at the lowest price, billed from a prepaid balance, and you pay only for generations that succeed.

Base URL: https://vidlab.ai/v1. The catalogue and estimates need no key; generations take an API key (Authorization: Bearer). Every page is Markdown: add .md to its docs URL.

## Quickstart

- [Overview](https://vidlab.ai/vidlab-api/docs/quickstart.md): One key, one request shape, every model: the base URL, auth and every endpoint on one page.
- [Authentication](https://vidlab.ai/vidlab-api/docs/authentication.md): API keys, how to send them and where to keep them.
- [Submit a generation](https://vidlab.ai/vidlab-api/docs/submit-a-generation.md): POST /v1/generations with max_cost_usd and an Idempotency-Key: submit, poll and download in cURL, Python and JavaScript.
- [Price it first](https://vidlab.ai/vidlab-api/docs/price-it-first.md): POST /v1/generations/quote prices the exact request for free, nothing held.
- [Get the result](https://vidlab.ai/vidlab-api/docs/get-the-result.md): Poll a generation, download the file with your key, and list your generations.

## Reference

- [Input fields](https://vidlab.ai/vidlab-api/docs/input-fields.md): Every input field, limits, bad-input answers, length from a file and an example for each kind of model.
- [Generation object](https://vidlab.ai/vidlab-api/docs/generation-object.md): Every field of a generation.
- [Statuses](https://vidlab.ai/vidlab-api/docs/statuses.md): queued, running, succeeded and failed, and what each means for your money.
- [Billing](https://vidlab.ai/vidlab-api/docs/billing.md): Prepaid credit, the reservation, pay only for success, reference-video billing, top-ups, auto-reload and refunds.
- [Balance](https://vidlab.ai/vidlab-api/docs/balance.md): GET /v1/balance: what your balance can spend now.
- [Uploads](https://vidlab.ai/vidlab-api/docs/uploads.md): POST /v1/uploads: send a file from your machine, accepted types and sizes, and the other ways to send a file.
- [Why a generation failed](https://vidlab.ai/vidlab-api/docs/why-a-generation-failed.md): The error object of a failed generation and every failure code. A failed generation is never charged.
- [Errors](https://vidlab.ai/vidlab-api/docs/errors.md): The error envelope, every request error code and how to fix each.
- [Model list](https://vidlab.ai/vidlab-api/docs/model-list.md): GET /v1/models: every model, its options, units, priced variants and status.

## Guides

- [Rate limits](https://vidlab.ai/vidlab-api/docs/rate-limits.md): Request limits, what does not count, Retry-After and safe retries with Idempotency-Key.
- [Formats](https://vidlab.ai/vidlab-api/docs/formats.md): Run our ad formats end to end: GET /v1/formats, start a run, answer its questions, poll for the finished video.
- [Viral videos](https://vidlab.ai/vidlab-api/docs/viral-videos.md): Search the viral video library over MCP or REST: the trending feed, filtered searches, one video in full, products and creators.

## For agents

- [MCP](https://vidlab.ai/vidlab-api/docs/mcp.md): Connect VidLab to Claude Code, Cursor, Codex, ChatGPT and Claude desktop.
- [Agent skill](https://vidlab.ai/vidlab-api/docs/agent-skill.md): A SKILL.md that teaches your agent to call the VidLab API.
- [llms.txt](https://vidlab.ai/vidlab-api/docs/llms-txt.md): Machine-readable docs for coding agents.

## Machine-readable

- [OpenAPI 3.1](https://vidlab.ai/v1/openapi.json): The /v1 API as JSON: every endpoint, input field, status and error.
- [llms-full.txt](https://vidlab.ai/vidlab-api/llms-full.txt): Every page above as one Markdown file.
- [SKILL.md](https://vidlab.ai/vidlab-api/SKILL.md): The Agent Skill: how an agent finds a model, prices it, runs it and fetches the files.

## Blog

- [All posts](https://vidlab.ai/blog): Prices, comparisons and setup guides for the VidLab API.
- [Cheapest AI video generation API in 2026: how to compare per-second pricing](https://vidlab.ai/blog/cheapest-ai-video-generation-api.md): A price per second only means something at one resolution, with audio on or off, for output seconds only. How to compare AI video API prices honestly, with a 720p price ladder from the VidLab catalogue.
- [Use Seedance 2.5 and Nano Banana from Claude, Cursor or Codex with the VidLab MCP (setup in one minute)](https://vidlab.ai/blog/vidlab-mcp-claude-cursor-codex.md): Connect Claude Code, Claude Desktop, Cursor or Codex to the VidLab MCP and let your assistant price and generate video and images with Seedance 2.5 and Nano Banana. The key stays in an environment variable.
- [Which AI video API should you use? Pick by task](https://vidlab.ai/blog/which-ai-video-api-should-you-use.md): A short guide to choosing an AI video model by what you need to make: drafts, social ads, talking heads, lip sync, video edits, 4K or upscaling. Models and per-second prices, all callable with one VidLab key.
- [Seedance 2.5 price per second: every resolution and mode](https://vidlab.ai/blog/seedance-2-5-price-per-second.md): Seedance 2.5 price per second on the VidLab API, from $0.06 a second: 480p, 720p and 1080p, the cost of a 5, 10, 15 or 30 second clip, what changes the price, and how it compares with other platforms.
- [UGC Real 1: a realistic AI UGC image model](https://vidlab.ai/blog/ugc-real-1-realistic-ai-ugc-image-model.md): UGC Real 1 is VidLab's own image model for realistic phone-shot UGC people. One flat price per image, words only, on the same API key as every other model. Examples, the price and a request.
- [Cheap Seedance 2.5 API with reference video: from $0.06 a second](https://vidlab.ai/blog/cheap-seedance-2-5-api-reference-video.md): Seedance 2.5 with a reference video on the VidLab API at $0.06, $0.15 and $0.34 a second, how the seconds are counted, and a worked example.
- [Sitemap](https://vidlab.ai/sitemap.xml): Every public page: blog posts, model pages and docs.