tutorial

Migrate Your API Integration to DeepSeek-V4 Before the July 2026 Deadline

Q: What prompt should I use?

// If using the Anthropic SDK with DeepSeek V4: const client = new Anthropic({ apiKey: '[YOUR_DEEPSEEK_API_KEY]', baseURL: 'https://api.deepseek.com', // Base URL remains unchanged }); const response = await client.messages.create({ model: 'deepseek-v4-pro', max_tokens: 1024, messages: [{ role: 'user', content: '[INSERT_PROMPT_HERE]' }], });

Published 06:03 MYT · Jul 1

Upgrade your API integrations to DeepSeek-V4 using OpenAI or Anthropic interfaces.

Switch your model parameters to 'deepseek-v4-pro' or 'deepseek-v4-flash' and utilize the new Anthropic interface support to modernize your API calls.

DeepSeek

The Scenario

You are managing a software project that uses AI for text generation and need to migrate to the latest model to ensure long-term stability before the July 2026 deprecation.

Before & after

The old way

Developers manually update API endpoints and request headers across multiple codebases, often taking 30–60 minutes per project whenever model versions shift.

With AI

Switching to the updated model names and using the Anthropic interface takes about 2–5 minutes of configuration.

The Prompt

// If using the Anthropic SDK with DeepSeek V4:
const client = new Anthropic({
  apiKey: '[YOUR_DEEPSEEK_API_KEY]',
  baseURL: 'https://api.deepseek.com', // Base URL remains unchanged
});

const response = await client.messages.create({
  model: 'deepseek-v4-pro',
  max_tokens: 1024,
  messages: [{ role: 'user', content: '[INSERT_PROMPT_HERE]' }],
});

The DeepSeek-V4 update introduces 'deepseek-v4-pro' and 'deepseek-v4-flash' while deprecating legacy names in July 2026. The new support for the Anthropic interface allows developers to swap DeepSeek into existing Claude-based workflows with minimal code changes.

Source

Change Log | DeepSeek API Docs

"The DeepSeek API now supports V4-Pro and V4-Flash, available via both the OpenAI ChatCompletions interface and the Anthropic interface."