Analyze and Describe Images with Meliai Vision Models

Vision-capable models can interpret images as part of their input, allowing you to build applications that describe photos, extract text from screenshots, audit diagrams, compare product images, and much more — all within a single API call alongside a text prompt. Meliai routes vision requests to European providers that support multimodal inference, keeping your image data within the EU throughout the entire request lifecycle. The interface follows the OpenAI multimodal message format, so any existing vision code works by changing only the base_url and API key.

Checking Vision Support

Before sending image content to a model, confirm it supports vision by fetching its metadata and inspecting the capabilities object.

curl

curl "https://api.meliai.ai/v1/models/<MODEL_ID>?include_meta=true" \
  -H "Authorization: Bearer $MELIAI_API_KEY"

Look for _meta.capabilities.vision: true in the response. Models that do not have this capability will reject requests containing image_url content parts.

Sending an Image

Replace the content string in a user message with an array of content parts. Include a text part for your prompt and an image_url part for each image you want the model to analyse.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["MELIAI_API_KEY"],
    base_url="https://api.meliai.ai/v1",
)

response = client.chat.completions.create(
    model="<MODEL_ID>",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What is shown in this image?"},
                {
                    "type": "image_url",
                    "image_url": {"url": "https://example.com/image.jpg"},
                },
            ],
        }
    ],
)

print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.MELIAI_API_KEY,
  baseURL: "https://api.meliai.ai/v1",
});

const response = await client.chat.completions.create({
  model: "<MODEL_ID>",
  messages: [
    {
      role: "user",
      content: [
        { type: "text", text: "What is shown in this image?" },
        {
          type: "image_url",
          image_url: { url: "https://example.com/image.jpg" },
        },
      ],
    },
  ],
});

console.log(response.choices[0].message.content);

curl https://api.meliai.ai/v1/chat/completions \
  -H "Authorization: Bearer $MELIAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "<MODEL_ID>",
    "messages": [
      {
        "role": "user",
        "content": [
          {"type": "text", "text": "What is shown in this image?"},
          {"type": "image_url", "image_url": {"url": "https://example.com/image.jpg"}}
        ]
      }
    ]
  }'

You can include multiple image_url parts in a single message to compare or jointly analyse several images.

Image Formats

Meliai accepts images in two forms:

Public URL
Base64 Data URI

Pass any publicly accessible HTTPS URL in the url field.

{
  "type": "image_url",
  "image_url": {
    "url": "https://example.com/image.jpg"
  }
}

Image URLs are fetched and re-encoded server-side by Meliai before the request is forwarded to the inference provider. The downstream provider never receives or stores your original URL, which preserves referrer privacy and ensures the image is processed within the EU.

Encode the image as a base64 data URI and pass it directly in the url field.

{
  "type": "image_url",
  "image_url": {
    "url": "data:image/jpeg;base64,/9j/4AAQSkZJRgAB..."
  }
}

Supported MIME types include image/jpeg, image/png, image/gif, and image/webp.

Use base64 data URIs for private or internal images that should not be served from a public URL. The image bytes travel directly in your API request payload and are never stored by Meliai after the request completes.

Multi-Turn Vision Conversations

You can reference images in earlier turns of a multi-turn conversation. Include the original image content part in the conversation history when sending follow-up messages — the model uses the image from the previous turn to answer questions about it in subsequent ones.

Python

messages = [
    {
        "role": "user",
        "content": [
            {"type": "text", "text": "Describe the chart in detail."},
            {"type": "image_url", "image_url": {"url": "https://example.com/chart.png"}},
        ],
    },
    {"role": "assistant", "content": "The chart shows quarterly revenue growth across..."},
    {"role": "user", "content": "Which quarter had the highest growth?"},
]

response = client.chat.completions.create(model="<MODEL_ID>", messages=messages)

Overview

Models & Routing

Features

Privacy

Analyze and Describe Images with Meliai Vision Models

Checking Vision Support

Sending an Image

Image Formats

Multi-Turn Vision Conversations

​Checking Vision Support

​Sending an Image

​Image Formats

​Multi-Turn Vision Conversations

Checking Vision Support

Sending an Image

Image Formats

Multi-Turn Vision Conversations