Developers / Docs / Vision

Vision

The Qwen models on ARK Cloud read images as well as text: documents, screenshots, charts, photos. Put images in the user message next to your question.

API reference: Chat completions

Models

Vision models have "ark_type": "multimodal" in /api/v1/models: arklabsai/Qwen3.6-35B-A3B, arklabsai/Qwen3.5-122B-A10B and arklabsai/Qwen3.5-4B. Send images only to these models.

Send an image

Use a content array with text and image_url parts. The URL can be a public https:// address, which ARK downloads, or a data: URL with base64 data.

import base64

with open("invoice.png", "rb") as f:
    b64 = base64.b64encode(f.read()).decode()

response = client.chat.completions.create(
    model="arklabsai/Qwen3.6-35B-A3B",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What is the total amount on this invoice?"},
            {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{b64}"}},
        ],
    }],
)
print(response.choices[0].message.content)

import { readFileSync } from "node:fs";

const b64 = readFileSync("invoice.png").toString("base64");
const response = await client.chat.completions.create({
  model: "arklabsai/Qwen3.6-35B-A3B",
  messages: [{
    role: "user",
    content: [
      { type: "text", text: "What is the total amount on this invoice?" },
      { type: "image_url", image_url: { url: `data:image/png;base64,${b64}` } },
    ],
  }],
});
console.log(response.choices[0].message.content);

curl https://api.ark-labs.cloud/api/v1/chat/completions \
  -H "Authorization: Bearer $ARK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "arklabsai/Qwen3.6-35B-A3B",
    "messages": [{
      "role": "user",
      "content": [
        {"type": "text", "text": "Describe this image."},
        {"type": "image_url", "image_url": {"url": "https://example.com/photo.jpg"}}
      ]
    }]
  }'

Several images can go in one message; refer to them in order ("the first image", "the second image").

Image size and tokens

Each image is resized to between 256 and 1,280 image tokens (about 0.26 to 1.3 megapixels), keeping its aspect ratio. Larger images do not cost more than the upper bound.
Image tokens are part of prompt_tokens. On ARK Cloud they also carry a small image surcharge; see Pricing.
The original images of one request may add up to 100 megapixels at most; larger requests get 400.
detail is accepted and has no effect.

Errors

An image that cannot be downloaded or decoded returns 400 with "The requested image cannot be processed".
Only text and image_url parts are accepted; input_audio and file parts return 400.

With the Anthropic API

Python

message = client.messages.create(
    model="arklabsai/Qwen3.5-122B-A10B",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": [
            {"type": "image", "source": {"type": "base64", "media_type": "image/png", "data": b64}},
            {"type": "text", "text": "What is the total amount on this invoice?"},
        ],
    }],
)