Vision - Yelu AI API

Vision-capable models can analyze images alongside text. In Chat Completions, send a user message whose content is an array containing text and image_url parts.

Analyze an image URL

Set IMAGE_URL to an HTTPS URL that the selected provider can fetch without authentication. For private content, prefer a short-lived signed URL or an inline data URI, subject to request-size limits.

curl https://api.yelu.ai/v1/chat/completions \
  -H "Authorization: Bearer $YELU_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{
    \"model\": \"gpt-4o-mini\",
    \"messages\": [{
      \"role\": \"user\",
      \"content\": [
        {\"type\": \"text\", \"text\": \"Describe the image and list any visible text.\"},
        {\"type\": \"image_url\", \"image_url\": {\"url\": \"$IMAGE_URL\", \"detail\": \"auto\"}}
      ]
    }],
    \"max_completion_tokens\": 300
  }"

const response = await fetch('https://api.yelu.ai/v1/chat/completions', {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${process.env.YELU_API_KEY}`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    model: 'gpt-4o-mini',
    messages: [{
      role: 'user',
      content: [
        { type: 'text', text: 'Describe the image and list any visible text.' },
        { type: 'image_url', image_url: { url: process.env.IMAGE_URL, detail: 'auto' } },
      ],
    }],
    max_completion_tokens: 300,
  }),
});
if (!response.ok) throw new Error(await response.text());
console.log((await response.json()).choices[0].message.content);

import os
import requests

response = requests.post(
    "https://api.yelu.ai/v1/chat/completions",
    headers={"Authorization": f"Bearer {os.environ['YELU_API_KEY']}"},
    json={
        "model": "gpt-4o-mini",
        "messages": [{
            "role": "user",
            "content": [
                {"type": "text", "text": "Describe the image and list any visible text."},
                {"type": "image_url", "image_url": {"url": os.environ["IMAGE_URL"], "detail": "auto"}},
            ],
        }],
        "max_completion_tokens": 300,
    },
    timeout=120,
)
response.raise_for_status()
print(response.json()["choices"][0]["message"]["content"])

Inline image data

Encode local images as a data: URL when the request stays within gateway and provider limits:

data:image/jpeg;base64,/9j/4AAQSkZJRgABAQ...

Use the correct media type, remove unnecessary metadata, resize images before encoding, and remember that base64 increases payload size by roughly one third.

Detail controls

When supported, detail can be:

low for faster, lower-cost coarse inspection;
high for detailed inspection and small text;
auto to let the model choose.

Reliability and privacy

Confirm the model supports vision before routing production traffic.
Validate media type, byte size, pixel dimensions, and URL scheme.
Block internal and link-local destinations if your server accepts user-supplied URLs.
Use short-lived signed URLs and least-privilege object access.
Remove sensitive metadata and redact images when required.
Do not rely on vision output alone for identity, medical, safety, or compliance decisions.
Ask for structured evidence such as observed text and regions, then validate it in application code.

Image input consumes model context and can be priced separately. Downscale to the resolution needed for the task and consult the account-specific pricing shown in the Yelu dashboard.

​Analyze an image URL

​Inline image data

​Detail controls

​Reliability and privacy

Analyze an image URL

Inline image data

Detail controls

Reliability and privacy