APIPod

Purpose

APIPod is an AI media generation API gateway for developers. Its goal is to use a single API key and a unified endpoint to call the generation capabilities of image, video, and language models, reducing the work of integrating separately with multiple model providers.

Main capabilities

• Unified access to multiple models: One endpoint covers video models such as Seedance, Veo, Sora, Wan, MiniMax, and Grok Imagine; image models such as Seedream, Nano Banana, and GPT Image; and large language models such as GPT, Claude, Gemini, Grok, and DeepSeek. The official page shows 145 models and 14 providers in Live status.
• Intelligent multi-channel routing: The same model can be configured with multiple backend channels, automatically selecting the optimal channel by weight or round robin to reduce costs and improve stability.
• Circuit breaker protection: After 3 consecutive failures, the circuit automatically breaks for 30 seconds, then probes and restores traffic to prevent cascading failures from affecting business.
• Automatic failover: When a retryable error occurs upstream, the request is forwarded to the next healthy channel.
• Observability: Provides real-time request logs, Token usage statistics, and a cost analysis dashboard.
• Flexible billing: Supports billing by Token, by request, and by duration; many models offer official, discounted, and special-price channels, allowing selection based on budget. Fees are reserved when a task is submitted, automatically refunded on failure, and each charge is listed item by item in the console.
• SDK compatibility: Chat completions are fully compatible with the OpenAI format; simply point the SDK's base_url to the API address shown in the console. It also supports the Anthropic Messages and native Gemini API formats.
• No-code use: The site includes built-in online AI image and video generators; enter a prompt to generate, sharing the same account and balance as the API; the prompt library provides ready-made prompts for each model.

How to use

1. Register and create an API key in the console.
2. Submit a generation task to obtain a task_id. For image tasks, call POST https://api.apipod.ai/v1/images/generations, and for video tasks call POST https://api.apipod.ai/v1/videos/generations.
3. Poll the corresponding status endpoint to get the result:GET https://api.apipod.ai/v1/images/status/{task_id} or GET https://api.apipod.ai/v1/videos/status/{task_id}. Tasks go through the pending, processing, completed, or failed states, and after completion you read the output URL from result. You can also pass callback_url, and the system will notify you when the task completes, so no polling is needed.

Official example code (Node.js):

const API = "https://api.apipod.ai/v1";
const headers = {
  Authorization: "Bearer " + process.env.API_KEY,
  "Content-Type": "application/json"
};

// 1. 提交生成任务
const { data } = await fetch(API + "/videos/generations", {
  method: "POST",
  headers,
  body: JSON.stringify({
    model: "seedance-2.5-t2v",
    prompt: "30-second cinematic aerial action spectacle: a battleship cuts through storm waves as a fighter jet bursts from volumetric clouds",
    aspect_ratio: "16:9",
    duration: 30
  })
}).then((r) => r.json());

// 2. 轮询直到任务完成
let task;
do {
  await new Promise((r) => setTimeout(r, 3000));
  const res = await fetch(API + "/videos/status/" + data.task_id, { headers });
  task = (await res.json()).data;
} while (!["completed", "failed"].includes(task.status));

console.log(task.result[0]); // → 输出 URL

Applicable Scenarios

• Autonomous AI agents: Chain multiple models (such as GPT-5.4 and Claude 4.6) to build agents that can reason, code, and execute tasks autonomously.
• Automated content production: Batch-generate blog posts, marketing copy, and social media images through a single API call pipeline.
• Enterprise AI gateway: Centralize all AI traffic into one gateway to uniformly enforce compliance, logging, and spending management.
• Online creation: Users who do not write code can directly use the on-site image and video generators and prompt library.

Limitations and Notes

• Images and videos are asynchronous tasks, and generation time depends on the model, resolution, and video length; the official fixed time values are not provided.
• The billing units and prices of each model vary by model. The page provides real-time rates (per 1 million tokens or the unit shown), and actual billing is subject to the console.
• Failed tasks are automatically refunded, but the cost is reserved first when the task is submitted.
• New accounts are provided with free trial credits and do not require a credit card; the specific credit amount is not currently provided.
• The input types, resolutions, and duration limits supported by each model vary by model. As listed on the page, Seedance 2.5 supports a maximum of 30 seconds per generation, native audio-video synchronization, and up to 30 images + 10 videos + 10 audio clips as references; Seedance 2.0 supports up to 4K 10-bit output and 4–15 second multi-shot sequences. Specific details are subject to each model's page.