Why Choose DeepSeek API

DeepSeek offers highly cost-effective large model API services, providing top-tier reasoning capabilities at prices far below similar products. Its API is compatible with the OpenAI format, making migration costs extremely low, making it the best choice for individual developers and small teams.

Registration and Getting API Key

After registering an account on the DeepSeek official website, create a new API Key on the "API Keys" page in the console. It is recommended to store the Key in environment variables to avoid hardcoding it in code:

# Linux/macOS
export DEEPSEEK_API_KEY="sk-your-api-key-here"

# Windows PowerShell
$env:DEEPSEEK_API_KEY="sk-your-api-key-here"

Installing SDK and First Call

pip install openai

from openai import OpenAI

client = OpenAI(
    api_key="sk-your-api-key",
    base_url="https://api.deepseek.com"
)

response = client.chat.completions.create(
    model="deepseek-chat",
    messages=[
        {"role": "system", "content": "You are a helpful AI assistant"},
        {"role": "user", "content": "Introduce DeepSeek in one sentence"}
    ],
    temperature=0.7,
    max_tokens=200
)

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

Model Selection Guide

DeepSeek offers two main models:

  • deepseek-chat: General conversational model, suitable for daily Q&A, text generation, translation, etc. Fast response, low cost
  • deepseek-reasoner: Deep reasoning model, suitable for complex tasks requiring deep thinking such as mathematics, programming, logical reasoning

Use deepseek-chat for simple tasks and deepseek-reasoner for complex reasoning, saving over 50% of costs.

Streaming Output Implementation

stream = client.chat.completions.create(
    model="deepseek-chat",
    messages=[{"role": "user", "content": "Write a poem about AI"}],
    stream=True
)

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

Context Management Best Practices

DeepSeek supports a 128K context window, but proper context management can significantly improve performance and reduce costs:

  • Multi-turn dialogue: Keep the most recent 5-10 turns, replace older ones with summaries
  • System prompt: Place it as the first message in the messages array, not counted in turn count
  • Token control: Use the tiktoken library to estimate token count and avoid exceeding limits

Error Handling and Retry

import time
from openai import RateLimitError, APIError

def call_with_retry(messages, max_retries=3):
    for attempt in range(max_retries):
        try:
            return client.chat.completions.create(
                model="deepseek-chat",
                messages=messages
            )
        except RateLimitError:
            wait = 2 ** attempt
            print(f"Rate limited, retrying in {wait}s...")
            time.sleep(wait)
        except APIError as e:
            if attempt == max_retries - 1:
                raise
            time.sleep(1)

result = call_with_retry([{"role": "user", "content": "Hello"}])

Next Steps

After mastering basic API calls, you can dive deeper into advanced features such as Function Calling, JSON mode output, streaming processing, and multi-turn dialogue management. The complete DeepSeek API documentation is available on the official website.