Vision

Vision models can analyze images alongside text prompts. Use them for image understanding, extraction, classification, visual question answering, and multimodal reasoning. Venice supports OpenAI-compatible multimodal chat messages. Put text and image blocks in the same user message, then send the request to a vision-capable model.

Basic Usage

import os
from openai import OpenAI

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

response = client.chat.completions.create(
    model="qwen3-vl-235b-a22b",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "Describe this image in three bullets."},
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "https://www.gstatic.com/webp/gallery/1.jpg"
                    },
                },
            ],
        }
    ],
)

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

import OpenAI from "openai";

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

const response = await client.chat.completions.create({
  model: "qwen3-vl-235b-a22b",
  messages: [
    {
      role: "user",
      content: [
        { type: "text", text: "Describe this image in three bullets." },
        {
          type: "image_url",
          image_url: {
            url: "https://www.gstatic.com/webp/gallery/1.jpg",
          },
        },
      ],
    },
  ],
});

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

curl https://api.venice.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3-vl-235b-a22b",
    "messages": [
      {
        "role": "user",
        "content": [
          {"type": "text", "text": "Describe this image in three bullets."},
          {
            "type": "image_url",
            "image_url": {
              "url": "https://www.gstatic.com/webp/gallery/1.jpg"
            }
          }
        ]
      }
    ]
  }'

Use Base64 Images

You can also pass a base64 data URL when the image is local or private:

{
  "type": "image_url",
  "image_url": {
    "url": "data:image/png;base64,<BASE64_ENCODED_IMAGE>"
  }
}

Replace <BASE64_ENCODED_IMAGE> with the complete base64 encoding of the image file, for example base64.b64encode(open("photo.png", "rb").read()).decode() in Python.

Base64 images must start with the data:image/<type>;base64, prefix (for example data:image/jpeg;base64,). Raw base64 without the prefix is rejected. Images must be at least 64×64 pixels, and SVG is not supported. If an image is rejected, the error message states the reason.

Send Multiple Images

content is an array, so one user message can include several image_url blocks alongside text. Use this to compare images or ask a question that spans more than one image:

{
  "model": "qwen3-vl-235b-a22b",
  "messages": [
    {
      "role": "user",
      "content": [
        {"type": "text", "text": "Compare these two images and list the differences."},
        {
          "type": "image_url",
          "image_url": {"url": "https://www.gstatic.com/webp/gallery/1.jpg"}
        },
        {
          "type": "image_url",
          "image_url": {"url": "https://www.gstatic.com/webp/gallery/2.jpg"}
        }
      ]
    }
  ]
}

Not every vision model accepts more than one image per request. Check supportsMultipleImages and maxImages under model_spec.capabilities in the Models API response before sending several. Each image also consumes context tokens. If a request with several images returns TOO_MANY_TOKENS (HTTP 400), send fewer images, shorten your instructions, or switch to a model with a larger context window.

Choose a Vision Model

Use the Text Models page or the Models API to find models that support vision. Vision support is listed in model capabilities.

For document-like inputs, use File Inputs when you want Venice to extract text from a file. Use vision when the visual layout or image content itself matters.

Prompting Tips

Tell the model what to focus on: objects, text, layout, safety, defects, or differences.
Ask for structured output when your application needs fields you can parse.
Keep image URLs accessible to the API, or use base64 data URLs for private images.
Use a model with enough context if you combine images with long instructions.

Docs

Getting Started

Text & Chat

Image

Video

Audio

Search & RAG

Agents & Integrations

Coding Tools

Agent Tooling

SDKs & Frameworks

Basic Usage

Use Base64 Images

Send Multiple Images

Choose a Vision Model

Prompting Tips

​Basic Usage

​Use Base64 Images

​Send Multiple Images

​Choose a Vision Model

​Prompting Tips

​Related Resources

Basic Usage

Use Base64 Images

Send Multiple Images

Choose a Vision Model

Prompting Tips

Related Resources