Skip to main content

DeepSeek

DeepSeek integration allows your product to connect with DeepSeek's API (an OpenAI-compatible API) to perform tasks like chat completion, reasoning, code generation, and more using DeepSeek models such as deepseek-chat or deepseek-reasoner.

Credentials Needed​

  • DeepSeek API Key — this is the bearer token you use in your HTTP requests.

There's no separate client secret, tenant ID, or OAuth flow currently documented — the API is keyed by a single token.

Permissions / API Access​

DeepSeek uses Bearer authentication; you must supply the API key in the Authorization: Bearer <API_KEY> header.

There are no scopes or granular permission sets documented; the API key gives access to whatever your DeepSeek account entitlements are.

Creating / Managing API Key​

  1. Sign up / log in to the DeepSeek API platform at their portal.
  2. Navigate to the API Keys section in your account dashboard.
  3. Create a new secret / API key. Optionally give it a name for your own reference.
  4. Copy the key once — you may not be able to view it again later.
  5. Optionally, in some setups, you may manage or revoke existing keys via that interface.

Test Connectivity​

Use curl or your preferred HTTP client to verify the key works:

curl https://api.deepseek.com/v1/chat/completions \
  -H "Authorization: Bearer <YOUR_DEEPSEEK_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
        "model": "deepseek-chat",
        "messages": [
          { "role": "user", "content": "Hello, DeepSeek!" }
        ]
      }'

If the request returns a valid JSON response with generated content, the connection is successful.

You can also adapt any OpenAI-compatible SDK by setting the base_url to https://api.deepseek.com and passing your DeepSeek API key in place of OpenAI's key.

Save the Results in the Platform and Create Connection​

  1. In your product's integration module, securely store the DEEPSEEK_API_KEY.
  2. Also store optional info like base_url (e.g. https://api.deepseek.com) and default model names (e.g. deepseek-chat, deepseek-reasoner).
  3. Create a connector entry (e.g., "DeepSeek Integration").
  4. As part of setup/validation, run a test call (like the above) to verify the key works and you can fetch a response.
  5. Capture any metadata you like (e.g., model names supported, version info) for your product's UI.

Best Practices​

  • Use the principle of least privilege: even though DeepSeek API keys don't appear to support scopes, limit usage of the key to the minimum models or endpoints you need.
  • Store API keys securely — in your secret vault, encrypted storage, not in version control.
  • Rotate API keys regularly and revoke keys that are no longer in use.
  • Monitor usage and errors to detect misuse or rate limit issues.
  • Use retries and backoff logic in the client when encountering transient errors (e.g., HTTP 5xx).
  • Allow base_url override so you can point to alternate endpoints or test environments.
  • Gracefully handle unauthorized (401) responses — e.g., notify user to check their key.
  • Cache static metadata or model info to avoid repeated API calls.
  • Log requests and responses (safely, without including sensitive content or full prompts) for debugging and auditing.