Completions API

Execute raw prompt completions without creating or storing a prompt first. Ideal for developers who want direct API access.

Authentication

Use your API key in the Authorization header:

Authorization: Bearer YOUR_API_KEY

Available Endpoints

Method	Endpoint	Description
POST	`/api/completions`	Main endpoint (use `?stream=true/false`)
POST	`/api/completions/sync`	Non-streaming response
POST	`/api/completions/stream`	HTTP/2 streaming (JSON array)
POST	`/api/completions/sse`	Server-Sent Events streaming

Request Body

{
  "promptText": "string (required)",
  "systemMessage": "string (optional)",
  "aiSystem": "string (required)",
  "model": "string (required)",
  "configuration": { },
  "replacements": { },
  "dataspaceIds": ["uuid"],
  "metadata": { }
}

Fields

Field	Type	Required	Description
`promptText`	string	Yes	The prompt text. Supports `{{variable}}` placeholders
`systemMessage`	string	No	System message for AI context
`aiSystem`	string	Yes	AI provider: `HyperleapAi` (included models), `OpenAi` or `Anthropic` (your own key)
`model`	string	Yes	Model id (e.g., `hl-gpt-6-luna`, `gpt-6-luna`, `claude-haiku-5-5`). See Available Models
`configuration`	object	No	Model parameters (`temperature`, `max_tokens`, etc.)
`replacements`	object	No	Key-value pairs for `{{variable}}` substitution
`dataspaceIds`	array	No	Dataspace UUIDs for RAG grounding
`metadata`	object	No	Custom metadata to attach

Examples

Basic Request

curl -X POST "https://api.hyperleapai.com/api/completions/sync" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "promptText": "Say hello in French",
    "aiSystem": "HyperleapAi",
    "model": "hl-gpt-6-luna"
  }'

With Variable Replacements

curl -X POST "https://api.hyperleapai.com/api/completions/sync" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "promptText": "Translate to {{language}}: {{text}}",
    "aiSystem": "HyperleapAi",
    "model": "hl-gpt-6-luna",
    "replacements": {
      "language": "Spanish",
      "text": "Hello world"
    }
  }'

With System Message

curl -X POST "https://api.hyperleapai.com/api/completions/sync" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "promptText": "What is the capital of France?",
    "systemMessage": "You are a geography expert. Be concise.",
    "aiSystem": "HyperleapAi",
    "model": "hl-gpt-6-luna"
  }'

Streaming (SSE)

curl -X POST "https://api.hyperleapai.com/api/completions/sse" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: text/event-stream" \
  -d '{
    "promptText": "Count from 1 to 5",
    "aiSystem": "HyperleapAi",
    "model": "hl-gpt-6-luna"
  }'

With Configuration

curl -X POST "https://api.hyperleapai.com/api/completions/sync" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "promptText": "Write a creative story about a robot",
    "aiSystem": "HyperleapAi",
    "model": "hl-gpt-6-luna",
    "configuration": {
      "max_tokens": 500
    }
  }'

Response Format

Sync Response

{
  "choices": [{
    "finish_reason": "Stop",
    "index": 0,
    "message": {
      "role": "Assistant",
      "content": "Bonjour!"
    }
  }],
  "messageId": "uuid",
  "runId": "uuid",
  "usage": {
    "prompt_tokens": 21,
    "completion_tokens": 7,
    "total_tokens": 28
  }
}

SSE Response

data: {"choice":{"index":0,"message":{"role":"assistant","content":"Hello"}},...}
 
data: {"choice":{"index":0,"message":{"role":"assistant","content":" world"}},...}
 
data: {"choice":{"index":0,"finish_reason":"Stop","message":{"role":"assistant"}},...}

Available Models

HyperleapAi

Included on every plan and billed in Hyperleap credits.

hl-gpt-6-luna (default)
hl-claude-haiku-5-5

OpenAi

Requires your own OpenAI key, set at the organization level (BYOK).

gpt-6-luna

Anthropic

Requires your own Anthropic key, set at the organization level (BYOK).

claude-opus-5-5
claude-sonnet-5-5
claude-haiku-5-5

GPT-6 models run at a fixed temperature of 1, and Claude 5.5 models don't use temperature or top_p. The API ignores those parameters for these models.

Error Handling

Status	Description
400	Invalid request (missing required fields, invalid model)
401	Unauthorized (invalid or missing API key)
500	Server error

Notes

Credits are consumed for each API call
Variable replacement uses Handlebars syntax: {{variableName}}
All endpoints support the same request body format
Use /api/completions?stream=true for streaming via query parameter

POST /api/completions
POST /api/completions/sync
POST /api/completions/stream
POST /api/completions/sse