SenseTime SenseNova API Developer Guide

What This Tutorial Solves

You will quickly get started with the SenseTime SenseNova large model:

• SenseChat text conversation
• Multimodal image-text understanding
• Function Calling
• Dual-mode: Python SDK / REST API

🎯 SenseTime is a global leader in computer vision. The SenseNova large model stands out especially in multimodal capabilities.

---

Model Family

Model	Capability	Context	Pricing
SenseChat 5.5	General text	128K	¥2/M
SenseNova-Vision	Image-text understanding	32K	¥3/M
SenseNova-Coder	Code generation	32K	¥2/M
SenseNova-TTS	Text-to-speech	-	¥2/10K chars

---

Quick Start

from openai import OpenAI
import os

client = OpenAI(
    api_key=os.getenv("SENSENOVA_API_KEY"),
    base_url="https://api.sensenova.cn/v1",
)

def chat(prompt: str, system: str = "") -> str:
    """Basic conversation"""
    messages = []
    if system:
        messages.append({"role": "system", "content": system})
    messages.append({"role": "user", "content": prompt})

    response = client.chat.completions.create(
        model="sensechat-5.5",
        messages=messages,
        temperature=0.7,
        max_tokens=2048,
    )
    return response.choices[0].message.content

# Test
result = chat("What are SenseTime's core technology strengths?")
print(result)

---

Multimodal Understanding

import base64

def vision_chat(image_path: str, question: str) -> str:
    """Image-text understanding"""
    with open(image_path, "rb") as f:
        b64 = base64.b64encode(f.read()).decode()

    response = client.chat.completions.create(
        model="sensenova-vision",
        messages=[
            {
                "role": "user",
                "content": [
                    {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{b64}"}},
                    {"type": "text", "text": question},
                ],
            }
        ],
        max_tokens=2048,
    )
    return response.choices[0].message.content

# Usage
description = vision_chat("street.jpg", "Describe the main buildings and human activity in this street scene photo")
print(description)

---

Function Calling

import json

# Define tools
tools = [
    {
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get weather information for a specified city",
            "parameters": {
                "type": "object",
                "properties": {
                    "city": {"type": "string", "description": "City name"}
                },
                "required": ["city"],
            },
        },
    }
]

def function_calling_demo(user_query: str):
    """Function calling demo"""
    response = client.chat.completions.create(
        model="sensechat-5.5",
        messages=[{"role": "user", "content": user_query}],
        tools=tools,
        tool_choice="auto",
    )

    msg = response.choices[0].message
    if msg.tool_calls:
        for tool_call in msg.tool_calls:
            func_name = tool_call.function.name
            args = json.loads(tool_call.function.arguments)
            print(f"Calling function: {func_name}({args})")

            # Execute the function
            if func_name == "get_weather":
                result = {"city": args["city"], "temp": 25, "weather": "Sunny"}
                print(f"Result: {result}")
    else:
        print(msg.content)

function_calling_demo("What's the weather in Beijing today?")

---

Streaming Chat

def stream_chat(prompt: str):
    """Streaming conversation"""
    stream = client.chat.completions.create(
        model="sensechat-5.5",
        messages=[{"role": "user", "content": prompt}],
        stream=True,
    )

    for chunk in stream:
        if chunk.choices[0].delta.content:
            print(chunk.choices[0].delta.content, end="", flush=True)

stream_chat("Write a short poem about AI")

---

Direct REST API Call

import requests

def rest_api_chat(prompt: str, api_key: str) -> str:
    """Direct REST API call"""
    response = requests.post(
        "https://api.sensenova.cn/v1/chat/completions",
        headers={
            "Authorization": f"Bearer {api_key}",
            "Content-Type": "application/json",
        },
        json={
            "model": "sensechat-5.5",
            "messages": [{"role": "user", "content": prompt}],
            "temperature": 0.7,
            "max_tokens": 2048,
        },
    )
    return response.json()["choices"][0]["message"]["content"]

---

SenseTime vs Other Models

Advantage	Description
Best Multimodal	Built on deep CV expertise, most thorough visual understanding
128K Context	SenseChat 5.5 supports ultra-long documents
Mature Industry Solutions	Vertical solutions for finance, healthcare, autonomous driving
Data Security	Supports private deployment

---

Next Steps

• Chinese Multimodal AI Guide
• Chinese AI Model Pricing Comparison

📝 Based on SenseTime SenseNova API, June 2026.