Endpoints

Vision & files

Send images, PDFs, audio and video as content parts in a chat message to api14 models, with supported formats, size limits and examples.

Try it in the playground

Multimodal models accept images, PDFs, audio and video in a user message. Pass content as an array of parts instead of a string. To see what a model accepts, check architecture.input_modalities in Models (image, file, audio, video).

Content parts

Part	Shape	Notes
Text	`{ "type": "text", "text": "..." }`
Image	`{ "type": "image_url", "image_url": { "url": "https://..." \| "data:image/png;base64,...", "detail": "auto" } }`	PNG, JPEG, WebP, GIF. URL or base64 data URL.
PDF / file	`{ "type": "file", "file": { "filename": "doc.pdf", "file_data": "https://..." \| "data:application/pdf;base64,..." } }`	Works with any model. Models without native file support get parsed text.
Audio	`{ "type": "input_audio", "input_audio": { "data": "<base64>", "format": "wav" } }`	Base64 only. `wav`, `mp3` and others, depending on the model.
Video	`{ "type": "video_url", "video_url": { "url": "https://..." } }`	Supported by some models, e.g. Gemini.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api14.com/api/v1",
  apiKey: process.env.API14_API_KEY,
});

const result = await client.chat.completions.create({
  "model": "openai/gpt-5-mini",
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "What is in this image?"
        },
        {
          "type": "image_url",
          "image_url": {
            "url": "https://upload.wikimedia.org/wikipedia/commons/3/3a/Cat03.jpg"
          }
        }
      ]
    }
  ]
});
console.log(result.choices[0].message.content);

Image example

{
  "model": "google/gemini-2.5-flash",
  "messages": [
    {
      "role": "user",
      "content": [
        { "type": "text", "text": "What is in this picture?" },
        { "type": "image_url", "image_url": { "url": "https://upload.wikimedia.org/wikipedia/commons/a/a7/Camponotus_flavomarginatus_ant.jpg" } }
      ]
    }
  ]
}

PDF example

{
  "model": "anthropic/claude-sonnet-4.5",
  "messages": [
    {
      "role": "user",
      "content": [
        { "type": "text", "text": "Summarize this contract in five bullets." },
        { "type": "file", "file": { "filename": "contract.pdf", "file_data": "data:application/pdf;base64,JVBERi0xLjcK..." } }
      ]
    }
  ],
  "plugins": [{ "id": "file-parser", "pdf": { "engine": "pdf-text" } }]
}

The optional file-parser plugin chooses how a PDF is read: pdf-text (free text extraction), mistral-ocr (OCR for scans, billed per page) or native (the model reads the file itself). The parsing cost is part of usage.cost.

Response

The response is a normal chat completion. Image, file and audio inputs count as input tokens, and their cost is in usage.cost:

{
  "choices": [{ "index": 0, "finish_reason": "stop", "message": { "role": "assistant", "content": "A close-up of an ant on a leaf." } }],
  "usage": { "prompt_tokens": 1290, "completion_tokens": 12, "total_tokens": 1302, "cost": 0.000417 }
}

Limits & errors

A request body can be at most 25 MB. Base64 makes files about 33% larger. Larger requests return 413 payload_too_large. For big images, use public URLs.
A model that doesn't accept an input type returns an upstream 400 or 404.
Other errors are the same as for Chat completions. See Errors.