> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bizyair.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Model API

> BizyAir Model Service: HTTP API for calling platform models with async tasks, WebHook callbacks, and CLI tools.

Model Service exposes the platform's model capabilities as HTTP APIs for external applications to call — built for developers and integration scenarios. It turns model capabilities into programmable interfaces, supporting asynchronous tasks, WebHook callbacks, file uploads, and other HTTP API operations, plus a CLI tool for batch management. Think of it as the service layer that converts "ComfyUI node capabilities → HTTPS API → business systems".

## Core Capabilities

| Capability          | Description                                                             |
| ------------------- | ----------------------------------------------------------------------- |
| Model API calls     | Invoke platform model capabilities via HTTPS POST                       |
| Async tasks         | Query results of long-running generation tasks through async interfaces |
| WebHook callbacks   | Proactively notify the caller via WebHook once a task completes         |
| File upload         | Upload images / models and other files as API inputs                    |
| CLI tool            | A command-line tool for batch management and invocation                 |
| Tag filter          | Quickly locate models via business tags                                 |
| Vendor filter       | Filter by vendor (Google / ByteDance / Alibaba / SiliconFlow …)         |
| Self-deployed badge | Open-source models with BizyAir-optimized deployment                    |

## Use Cases

* **Backend integration**: Bring BizyAir's model capabilities into your own product backend
* **Agent invocation**: Let AI agents call models via API to generate images / videos
* **Batch processing**: Submit generation tasks in bulk via CLI or scripts
* **Service monetization**: Wrap model capabilities into an outward-facing SaaS service

## Capability Categories

The Model Service APIs are divided into 16 capability categories along the "input → output" dimension.

### Category Distribution

| Category                  | Category      | Category       | Category       |
| ------------------------- | ------------- | -------------- | -------------- |
| Text to Video             | Text to Image | Image to Image | Image to Video |
| First-Last Frame to Video | Video Editing | Audio to Audio | Text to Speech |
| Reference to Video        | LLM           | VLM            | Text to Audio  |
| Video Extension           | 3D Generation | Speech to Text | Text to Lyrics |

## Vendor Distribution

More vendors are being onboarded.

| Vendor        | Vendor              | Vendor            | Vendor            |
| ------------- | ------------------- | ----------------- | ----------------- |
| Google        | ByteDance           | Alibaba           | SiliconFlow       |
| Kunlun Wanwei | Shengshu Technology | Kuaishou / Kling  | XAI / grok        |
| OpenAI        | MiniMax (Hailuo)    | Black Forest Labs | Jimeng (Dreamina) |

> **Note:** `owner=Self` denotes a SiliconFlow self-deployed model; `owner=Third` denotes a third-party channel version. Billing rates may differ.

## API Model Field Reference

Every API model includes the following core fields:

| Field         | Description                                                   |
| ------------- | ------------------------------------------------------------- |
| display\_name | Model display name                                            |
| description   | Description of the model's capabilities                       |
| category      | Capability category (e.g., `Text to Image`, `Image to Video`) |
| model\_name   | Internal model name (used in API calls)                       |
| endpoint      | API call endpoint path                                        |
| tags          | Array of business tags                                        |
| status        | Status (`normal` / `gray` for gray release)                   |
| gray          | Whether in gray release                                       |
| new\_tag      | Whether newly launched                                        |
| owner         | Source (`Self` self-deployed / `Third` third-party)           |
| manufacturer  | Providing vendor                                              |

## How to Call

### Direct HTTP Call

1. Find the target API model on the Model Service page.
2. Get the `endpoint` and request body format from the `API Reference` docs.
3. Include `Authorization: Bearer <token>` in the request header (get the token from your personal center).
4. POST the request; the response returns a `requestId` for subsequent queries.

### Async Tasks

1. Keep the returned `requestId` after submitting a task.
2. Poll the task status via the async query interface.
3. Retrieve the result URL once the task completes.

### WebHook Callback

1. Set `webhook_url` in the request parameters.
2. Once the task completes, the platform POSTs the callback result to that URL.
3. Verify the signature server-side, then persist the result.

### CLI Tool

1. Install the BizyAir CLI tool.
2. Log in to your account via `bizyair login`.
3. Invoke a model with `bizyair run <endpoint> --input <file>`.
4. Query task status with `bizyair task get <requestId>`.

> **Note:** For detailed API fields, error codes, and CLI commands, see the `API Reference`, `Return Code Reference`, and `CLI Tool` chapters at docs.bizyair.vip.

## Billing and Pricing

For model pricing, see [Node Pricing](/pricing/node-price).