VideoLive

Wan 3 Spicy

Create videos from a prompt, keyframes or visual references, with optional audio.

Output / starting price

$0.0385/ second

Choose by workload—not by the integration you already wrote.

Compare inputs, outputs, pricing, and regional availability through one interface, then pick the right model for every job.

Model series

Wan 3 Spicy

Live

modelrush/wan3-spicyCreate videos from a prompt, keyframes or visual references, with optional audio.

Output / starting price$0.0385

/ second· 480P

View model

Integrate directly—or hand the job to your agent.

Let your agent integrate ModelRush

Copy one complete implementation brief so an engineering agent can inspect, edit, and verify the project.

Copy prompt

Integrate ModelRush into this project. Inspect the stack first, install the appropriate OpenAI SDK, read MODELRUSH_API_KEY from a server-side environment variable, set the base URL to https://api.modelrush.ai/v1, preserve request_id, add timeout and retry handling, then run one verification request with modelrush/grok-4.5.

Open the agent integration guide

Manual integration

Use a familiar OpenAI SDK in TypeScript or Python.

Copy code
import OpenAI from "openai";
const client = new OpenAI({
  apiKey: process.env.MODELRUSH_API_KEY,
  baseURL: "https://api.modelrush.ai/v1",
});
const result = await client.chat.completions.create({
  model: "modelrush/grok-4.5",
  messages: [{ role: "user", content: "Draft a product launch." }],
});
console.log(result.id);

View full documentation

Production-ready from the first request

One API keyUse the same credentials for text, image, video, and audio.

Unified request tracingPreserve request_id, status, latency, and errors for debugging.

Regional routingChoose an execution path based on model and regional availability.

Agent-readyCopy a complete prompt and let an engineering agent integrate and verify it.

Infrastructure handles the complexity. Your team ships the product.

One request contract

Keep product code independent from one model vendor and change models without rewriting the interface.

Every call is traceable

Request ID, model, status, tokens, latency, and errors land in one record.

Selectable regions

Choose the right region for each workload and preserve the actual execution path.

Clear data boundaries

Keep keys server-side and make provider differences and data practices visible at the model layer.

Latest from the blog

Workflow

New in ModelRush: AI Writing, Storyboards, and Batch Face Swap

Read article

Video models

Wan 3.0 Price Cut: Lower Standard and Pro Rates, New Prime Spicy Models

Read article

Agents

How to Build Your Own AI Girlfriend with ModelRush

Read article

Common questions before the first integration.

ModelRush is a unified model infrastructure layer. Your product keeps one request contract while the execution layer can connect different models and regions.

Yes. Set the base URL to https://api.modelrush.ai/v1 and select a live ModelRush model ID.

Yes. Image generation and editing, video generation, speech transcription, and text-to-speech endpoints are connected.

Every request gets a unique ID and records its model, status, tokens, latency, and errors for dashboard inspection.

Provider evaluation is still in progress. Current public pricing and billing status are shown on the pricing page and in the dashboard.

Turn multi-model access into infrastructure your product can keep building on.

Talk with us about model selection, regional deployment, usage, data boundaries, and commercial plans.

Contact sales

ModelRushOne integration, intelligent routing, transparent billing. Model infrastructure for developers and agents.

ProductModels Pricing Dashboard Blog

DevelopersDocs Agent integration DIY AI Girlfriend

Privacy & Terms Acceptable Use Data Retention Data Deletion Policy

GitHub

Hugging Face