API request examples
If you use a coding tool, start with Tool setup. This page is for developers sending HTTP requests directly.
These examples use Bash or Zsh on macOS / Linux. Set KEENKIND_API_KEY, then replace YOUR_MODEL_ID with a model from the catalog confirmed for the chosen API.
export KEENKIND_API_KEY='YOUR_KEENKIND_API_KEY'Choose the request you need; you do not need to run all three. Generation requests incur actual usage.
Responses
Jump to Responsescurl --silent --show-error 'https://api.keenkind.ai/v1/responses' \ --header "Authorization: Bearer $KEENKIND_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ "model": "YOUR_MODEL_ID", "input": "Explain unit testing in two sentences." }'Chat Completions
Jump to Chat Completionscurl --silent --show-error 'https://api.keenkind.ai/v1/chat/completions' \ --header "Authorization: Bearer $KEENKIND_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ "model": "YOUR_MODEL_ID", "messages": [ {"role": "user", "content": "Explain unit testing in two sentences."} ] }'Anthropic Messages
Jump to Anthropic Messagescurl --silent --show-error 'https://api.keenkind.ai/v1/messages' \ --header "x-api-key: $KEENKIND_API_KEY" \ --header 'anthropic-version: 2023-06-01' \ --header 'Content-Type: application/json' \ --data '{ "model": "YOUR_MODEL_ID", "max_tokens": 512, "messages": [ {"role": "user", "content": "Explain unit testing in two sentences."} ] }'Use an SDK
Jump to Use an SDKConfigure the matching protocol’s SDK with:
| SDK | Base URL | Key |
|---|---|---|
| OpenAI-compatible SDK | https://api.keenkind.ai/v1 |
Value of KEENKIND_API_KEY |
| Anthropic-compatible SDK | https://api.keenkind.ai |
Value of KEENKIND_API_KEY |
Use an id from the KeenKind catalog. Third-party integrations do not need website login cookies or website-client-specific headers.
Check the result
Jump to Check the resultA successful request returns a model response in the selected protocol, with a matching entry in usage records. For errors, check the status and error code in API troubleshooting. After a network interruption, review records before retrying.