Skip to main content
Attachments let you send file inputs (images, audio, PDFs, Markdown, …) to the Gateway. They can either be embedded as binary data or referenced with an URL.
Attachment support varies by provider and model version. The Gateway forwards attachments directly to the provider, so unsupported types will return a provider error.Many modern models support images, video, audio, and PDFs; some also support Office formats (e.g., DOCX, PPTX, XLSX). Refer to /models and the provider’s official documentation for the latest capability details.

​
From a URL (e.g. Markdown page)

If you have a file hosted on a URL, you can reference it directly in the message. The Vercel AI SDK will download it and then attach the binary to the message. 
import { createOpenAICompatible } from "@ai-sdk/openai-compatible";
import { generateText } from "ai";

const hebo = createOpenAICompatible({
  name: "hebo",
  apiKey: process.env.HEBO_API_KEY,
  baseURL: "https://gateway.hebo.ai/v1",
});

const { text } = await generateText({
  model: hebo("openai/gpt-oss-20b"),
  messages: [
    {
      role: "user",
      content: [
        {
          type: "text",
          text: "How to find out which models are supported by Hebo?",
        },
        {
          type: "file",
          data: new URL("https://hebo.ai/docs/gateway/supported-models.md"),
          mediaType: "text/markdown",
        },
      ],
    },
  ],
});

console.log(text);

​
From local File (e.g. Image)

If you already have a file locally, you can attach the binary data to the model call. Internally, it will automatically be converted to base64. 
import { createOpenAICompatible } from "@ai-sdk/openai-compatible";
import { generateText } from "ai";
import { readFile } from "node:fs/promises";

const hebo = createOpenAICompatible({
  name: "hebo",
  apiKey: process.env.HEBO_API_KEY,
  baseURL: "https://gateway.hebo.ai/v1",
});

const { text } = await generateText({
  model: hebo("google/gemini-3-flash-preview"),
  messages: [
    {
      role: "user",
      content: [
        {
          type: "text",
          text: "Describe this image.",
        },
        {
          type: "image",
          image: await readFile("<FILENAME>"),
        },
      ],
    },
  ],
});

console.log(text);