How to beginner · 3 min read

How to fine-tune DeepSeek model

Quick answer

DeepSeek models do not currently support user-accessible fine-tuning via their API. Instead, customize behavior by prompt engineering or using few-shot learning with deepseek-chat. For advanced use, contact DeepSeek for enterprise fine-tuning options.

PREREQUISITES

Python 3.8+
DeepSeek API key
pip install openai>=1.0

Setup

Install the openai Python package to interact with DeepSeek's OpenAI-compatible API. Set your DeepSeek API key as an environment variable for secure authentication.

bash

pip install openai>=1.0

Step by step

DeepSeek does not provide public fine-tuning endpoints. Instead, use prompt engineering or few-shot examples with the deepseek-chat model to customize outputs. Below is a sample Python script demonstrating this approach.

python

import os
from openai import OpenAI

client = OpenAI(api_key=os.environ["DEEPSEEK_API_KEY"], base_url="https://api.deepseek.com")

messages = [
    {"role": "system", "content": "You are a helpful assistant specialized in summarization."},
    {"role": "user", "content": "Summarize the following text briefly: DeepSeek is a leading AI company."}
]

response = client.chat.completions.create(
    model="deepseek-chat",
    messages=messages
)

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

output

DeepSeek is a top AI company known for advanced language models.

Common variations

You can adjust the prompt style or add few-shot examples in the messages array to guide the model's behavior. DeepSeek currently does not support asynchronous or streaming fine-tuning calls. For different tasks, switch the model parameter to other DeepSeek models like deepseek-reasoner.

Troubleshooting

If you receive authentication errors, verify your DEEPSEEK_API_KEY environment variable is set correctly.
If the model output is not as expected, refine your prompt or add more few-shot examples.
For fine-tuning requests, note that DeepSeek does not currently expose this feature publicly; contact support for enterprise options.

✅

Key Takeaways

DeepSeek models do not support public fine-tuning via API; use prompt engineering instead.
Use the OpenAI-compatible deepseek-chat model with few-shot examples to customize output.
Set your API key in DEEPSEEK_API_KEY and use the openai SDK with base_url pointing to DeepSeek.
For advanced fine-tuning, contact DeepSeek enterprise support.
Adjust prompts iteratively to improve model responses without fine-tuning.

Verified 2026-04 · deepseek-chat, deepseek-reasoner

Verify ↗