Create Infinite Context Response | Captain API Documentation

Respond to an input prompt with Captain’s infinite context window management.

Under the hood, Captain splits an input into 30-40% context chunks (for good accuracy). LLMs process these chunks in parallel, and the responses are generatively merged for a comprehensive output.

Headers:

Authorization: Bearer {api_key} - Captain API key for authentication
X-Organization-ID: Organization UUID

Parameters:

input: Natural language text content (can be very large, 100M+ tokens)
query: Optional specific question to answer from the input (if not provided, generates overview)
stream: Enable real-time streaming via S2 (default: False)

Returns:

output: Natural language response to the input

Respond to an input prompt with Captain's infinite context window management. Under the hood, Captain splits an input into 30-40% context chunks (for good accuracy). LLMs process these chunks in parallel, and the responses are generatively merged for a comprehensive output. Headers: - Authorization: Bearer {api_key} - Captain API key for authentication - X-Organization-ID: Organization UUID Parameters: - input: Natural language text content (can be very large, 100M+ tokens) - query: Optional specific question to answer from the input (if not provided, generates overview) - stream: Enable real-time streaming via S2 (default: False) Returns: - output: Natural language response to the input

Request

This endpoint expects an object.

inputstringRequired

querystring or nullOptional

streamboolean or nullOptionalDefaults to false

toolsstring or nullOptional

tool_choicestring or nullOptional

Response

Successful Response

1	curl -X POST https://api.runcaptain.com/v1/responses \
2	-H "Content-Type: application/x-www-form-urlencoded" \
3	-d "input=Your%20extremely%20large%20text%20content%20here...%20This%20can%20be%20millions%20of%20tokens%20long." \
4	-d "query=What%20are%20the%20key%20themes%20in%20this%20document%3F" \
5	-d "stream=true"

Request

Response

Errors