Six REST endpoints
/api/v1/models, /api/v1/effects, /api/v1/prompts, /api/v1/generations, /api/v1/generations/:id and /api/v1/me cover model discovery, one-photo effects, prompt search, task creation, task status and your account.
Developers
Generate images and videos with every top model, run one-click effects and search thousands of proven prompts — from your code or your AI agent.
Create an API key in Settings → API Keys and send it with every request that costs credits or reads your generations.
Authorization: Bearer <YOUR_API_KEY>/api/v1/modelsModels, parameters and starting credits/api/v1/effectsOne-click effects and product recipes/api/v1/prompts?q=&type=&model=&limit=Search the open prompt library/api/v1/generationsStart a generation (model or effect) · API key/api/v1/generations/:idTask status and result URLs · API key/api/v1/generationsYour recent generations · API key/api/v1/meCredit balance · API keyGenerations run asynchronously. Create a task, then poll it every few seconds until the status is succeeded or failed. Failed tasks are refunded automatically.
curl -X POST https://neomundo.art/api/v1/generations \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2",
"prompt": "Slow dolly-in on a perfume bottle as petals fall, soft morning light",
"aspect_ratio": "9:16",
"resolution": "720p",
"duration": 5
}'
# => { "data": { "id": "…", "status": "processing", "credits": 198, … } }
curl https://neomundo.art/api/v1/generations/<id> -H "Authorization: Bearer $API_KEY"
# => { "data": { "status": "succeeded", "results": [{ "url": "…", "type": "video" }] } }Or run an effect / product recipe with just an image:
curl -X POST https://neomundo.art/api/v1/generations \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{ "effect": "product-white-bg", "images": ["https://…/product.jpg"] }'Connect Claude, Cursor or any MCP client and let your agent search prompts, pick models and generate media for you.
Claude Code
claude mcp add --transport http neomundo https://neomundo.art/api/mcp \
--header "Authorization: Bearer <YOUR_API_KEY>"Cursor / other clients
{
"mcpServers": {
"neomundo": {
"url": "https://neomundo.art/api/mcp",
"headers": { "Authorization": "Bearer <YOUR_API_KEY>" }
}
}
}Available tools:
search_prompts search the open prompt library
list_models models and parameters
list_effects effects and product recipes
generate start a generation
run_effect run an effect on images
get_generation check a task
get_credits credit balanceSearch community prompts by keyword, type and model. No key required. Each result includes the original author and source link — please keep the attribution when you reuse a prompt.
curl "https://neomundo.art/api/v1/prompts?q=product+ad&type=video&limit=5"FAQ
Create a key in Settings → API Keys and send it in the "Authorization: Bearer" header on every request. The open prompt library endpoint works without a key.
API generations use the same credits and per-model prices as the website, shown by /api/v1/models. Failed generations are refunded automatically. New accounts get 50 free credits on sign-up.
Yes. Content you generate can be used in ads, stores and social media, subject to the terms of service and the model providers' policies. Prompts from the library should keep their author attribution.
No. You create a task and poll /api/v1/generations/:id. Images usually finish in 10–60 seconds and videos in 1–5 minutes, depending on model and queue.
Up to 8 generations per account can run at the same time, whether they come from the API, the MCP server or the website.
Claude Code, Cursor and other clients that support MCP. Point them at /api/mcp with your API key to use the seven tools.
Yes. The same endpoints cover Nano Banana 2, GPT Image 2 and Seedream 5 Lite for images, as well as motion, swap, edit and avatar models.
Sign up, create a key in Settings → API Keys and make your first call to /api/v1/models. New accounts start with 50 free credits to test real generations.
Open Developer Docs/api/v1/models, /api/v1/effects, /api/v1/prompts, /api/v1/generations, /api/v1/generations/:id and /api/v1/me cover model discovery, one-photo effects, prompt search, task creation, task status and your account.
Create a key in Settings → API Keys and send it in the "Authorization: Bearer" header. No separate developer account or billing setup is needed.
A generation request returns a task ID right away. Poll /api/v1/generations/:id until the result is ready; images usually take 10–60 seconds and videos usually 1–5 minutes.
API calls draw from the same credit balance as the website at the same per-model prices, and failed generations are refunded automatically.
Run up to 8 generations at the same time per account, enough to render variations, several aspect ratios or a small batch in parallel.
Seven tools, search_prompts, list_models, list_effects, generate, run_effect, get_generation and get_credits, let Claude Code, Cursor and other MCP clients create media directly.
An AI video generation API lets your own code or agent send a prompt and receive a finished video, without opening a web studio. NeoMundo exposes the same 13 models that power its website through a REST API: video models such as Seedance 2.0 by ByteDance, Veo 3.1 by Google, Kling 3.0 by Kuaishou and Hailuo H3 by MiniMax, and image models such as Nano Banana 2, GPT Image 2 and Seedream 5 Lite, so it also works as an AI image generation API.
The flow is asynchronous because video takes time. You call /api/v1/models to see each model's inputs and parameters, create a task with /api/v1/generations, then poll /api/v1/generations/:id for its status and result. /api/v1/effects lists the one-photo effects you can run with a single image, and /api/v1/me returns your account details, including your credit balance.
For AI agents there is an MCP server at /api/mcp. It gives Claude Code, Cursor and any other MCP client seven tools: search_prompts to find prompts in the public gallery, list_models and list_effects to discover what is available, generate and run_effect to start work, get_generation to check a task and get_credits to see the balance. An agent can go from idea to finished clip in one conversation.
Pricing is shared with the website. Every model has the same credit cost through the API as in the studio, starting at 9 credits for Seedream 5 Lite images and 53 credits for Seedance 2.0 Fast videos, and failed tasks are refunded automatically. The prompt library endpoint is open and needs no key; it returns each prompt with its author attribution so you can credit creators in your product.
Sign in, open Settings → API Keys and create a key. Store it as a secret; it spends credits from your account.
Call /api/v1/models with the header "Authorization: Bearer YOUR_KEY" to read each model's inputs, aspect ratios, resolutions, durations and credit cost.
Send your prompt, model and settings to /api/v1/generations, or run a preset from /api/v1/effects with one photo. The response includes a task ID.
Check /api/v1/generations/:id until the task is finished, then fetch the image or video. Keep no more than 8 tasks running at once.
Generate product photos with the AI image generation API and turn them into 9:16 product videos for every item in a catalog, running up to 8 tasks in parallel.
Connect the MCP server to Claude Code or Cursor and ask for a clip in plain language; the agent picks a model, calls generate and checks get_generation for you.
Use the open prompt library endpoint to show real prompts with author attribution inside your own tool, with no key required.
Schedule a script that picks a prompt, runs a one-photo effect or a short video and posts the result, checking get_credits before each batch.