DeepSeek / Developer guide
Start with the
DeepSeek API.
Find the official platform, choose a model identifier, and keep your integration separate from the consumer chat app.
1. Start at the official developer platform.
Open platform.deepseek.com to manage your developer account and API credentials. Review billing and available access before sending requests. Your official chat session is not an API key.
2. Choose the model you intend to call.
Use the current API identifier, rather than guessing it from a release name. Our model reference shows the current mappings; confirm them against the provider documentation before deployment.
3. Configure the connection on your server.
DeepSeek documents OpenAI-compatible and Anthropic-compatible request formats. Select the one that matches your integration and follow its official setup instructions.
| Setting | Official value |
|---|---|
| OpenAI-compatible base URL | https://api.deepseek.com |
| Anthropic-compatible base URL | https://api.deepseek.com/anthropic |
| API model identifiers | deepseek-flash, deepseek-v4-pro |
Keep the API key on your server. Store it in a protected environment setting. Do not embed it in browser JavaScript, screenshots or a public repository.
4. Make a small, controlled first request.
Start with a short prompt and a modest response limit. Confirm which model handled it, inspect returned usage information, and compare that usage with the current API rates. Then test the error cases and limits your own application must handle.
What the API does not automatically include
An API integration is a separate application. Search, file handling, conversation storage and other tools depend on what you implement and what the chosen endpoint supports. Access to a model does not reproduce every feature of the official chat app.
Sources: DeepSeek API introduction · Developer platform. Checked September 18, 2026.