ModelRush is a hosted multi-model API service for developers, covering text, images, video and voice. It is a candidate for adding model capabilities to existing applications. Chat uses an OpenAI-compatible format; media integration follows each model's documentation. Sharing an account and API key does not mean every endpoint shares its inputs, outputs or task states.
Available capabilities
The documentation lists chat completions, image generation and editing, video generation and status queries, speech synthesis and audio transcription, plus media uploads and webhooks. The catalog shows model IDs, capabilities, regions and current prices. Select a currently available model rather than assuming a brand on a marketing page is accessible to your account.
Screenshot of the public documentation, showing endpoint navigation and status guidance.
Live, Beta and Planned distinguish available capabilities, contracts that may change and future plans. CLI, MCP and Agent skills navigation entries carry Planned labels; those pages do not establish that usable integration tools have shipped. Check endpoint status and account permissions before launch.
Chat compatibility and media schemas
The API base URL is https://api.modelrush.ai/v1. Chat uses POST /chat/completions; the official example configures the OpenAI SDK with a server-side key and this base URL. Obtain model IDs from the current catalog instead of treating listed names as permanent availability guarantees.
Image, video and voice endpoints have their own request and response structures. A single generic OpenAI media implementation will not cover them. Asynchronous jobs also require submission, waiting, success and failure handling. Follow the returned polling URL and its documented states rather than mixing status names across endpoints.
Getting started
- Choose a model and check its inputs, outputs, regions and billing unit.
- Create an API key after signing in, check balance and permissions, and keep the key server-side.
- Send a small request, retain its request ID and compare the result with billing records and errors.
- Implement media polling or webhooks separately. Handle timeouts and retryable errors according to the documentation, avoiding duplicate charges from repeated submissions.
- Set application budgets, concurrency and usage limits before increasing traffic.
Our editorial advice is to validate a real task before choosing by catalog size. Output acceptance, failure handling and cost per accepted result all affect the decision.
API credits and Agent subscriptions
Model API calls consume prepaid credits according to actual usage. Models use different billing units; check current prices and checkout methods on the pricing page and in your account. One blanket multiplier cannot describe every model, and a starting price does not cover every resolution or generation setting.
The website separately offers monthly Agent subscriptions for personal Agent runtime and storage, including an Agent Basic trial for eligible new subscribers. Its pricing page states that model API usage is excluded from membership fees and billed separately from credits. Evaluate the API as a paid service; the Agent trial is not an API free tier.
The billing guide points to pricing, the model catalog and the authenticated Billing dashboard for prices, availability and account usage respectively. Retain request IDs and reconcile costs by model, feature and accepted result rather than looking only at top-up totals.
Limits and selection
The catalog includes Spicy models requiring adult-access confirmation on the account. Applications serving minors or enforcing a defined content scope should assess whether the catalog and policies fit their requirements. Confirm rights to uploaded media and review data-handling terms.
Models and prices can change. We have no independent latency, stability or SLA verification, and have not confirmed official model-provider authorization. Mainland China connectivity, registration and payment have not been tested; access to public pages does not establish mainland availability. The submitter supplied info@modelrush.ai as its support address.
For multi-provider language-model routing, also consider OpenRouter. For managing application workflows and knowledge bases yourself, consider Dify. Its application-building role differs from a hosted model API, so first decide whether you need a calling service or an application platform.