Vision (image input)

Send images via OpenAI-standard image_url to multimodal models. SeaLink translates to each upstream's native format.

Models with image input

claude-sonnet-4-6 · claude-haiku-4-5 · claude-opus-4-7
gpt-4o · gpt-4o-mini
qwen3-max · qwen3-plus
doubao-1-5-pro
glm-4-6

Example

Both URL and base64 work. Use base64 for local / private images — but the request body grows (mind your token count).

Python

from openai import OpenAI
import base64
import { apiBaseUrl } from "@/lib/api-base";

client = OpenAI(
    base_url="https://api.sealink.asia/v1",
    api_key="<your-sealink-key>",
)

# From a URL:
resp = client.chat.completions.create(
    model="claude-sonnet-4-6",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What's in this image?"},
                {
                    "type": "image_url",
                    "image_url": {"url": "https://example.com/photo.jpg"},
                },
            ],
        }
    ],
)
print(resp.choices[0].message.content)

# From a local file (base64):
with open("invoice.png", "rb") as f:
    b64 = base64.b64encode(f.read()).decode()

resp = client.chat.completions.create(
    model="qwen3-max",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "Extract the total amount and date."},
                {
                    "type": "image_url",
                    "image_url": {"url": f"data:image/png;base64,{b64}"},
                },
            ],
        }
    ],
)

Typical use cases

Invoice / receipt OCR with field extraction
E-commerce product attribute extraction (color / material / style)
UI screenshot description / automated test assertions
Student homework / exam grading
Screenshare + agent control

Token billing

Each model tokenizes images differently (Claude charges by image size; OpenAI charges by tile count). See per-model rules on each model detail page.