Catalog/llm-self-consistency

LLM

Self-consistency vote API

Self-consistency / majority vote over N candidate answers (case- and whitespace-normalized): picks the modal answer, reports vote share as confidence, flags full agreement, and returns the full tally — implements the self-consistency sampling technique. Answers 'which answer do most of my samples agree on?', 'how confident is the majority vote?'.

Price$0.03per request
MethodPOST
Route/v1/llm/self-consistency
StatusLive
MIME typeapplication/json
Rate limit120/minute
Cache0s public
llmself-consistencyvotemajorityensemblesamplingconfidenceagent
API URLhttps://x402.hexl.dev/v1/llm/self-consistency
Integration docs
Example request
{
  "candidates": [
    "42",
    "42",
    "41",
    "42"
  ]
}
Example response
{
  "winner": "42",
  "votes": 3,
  "total": 4,
  "confidence": 0.75,
  "agreement": false,
  "tally": [
    {
      "answer": "42",
      "count": 3
    },
    {
      "answer": "41",
      "count": 1
    }
  ]
}
Input schema
{
  "type": "object",
  "required": [
    "candidates"
  ],
  "properties": {
    "candidates": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "examples": [
        [
          "42",
          "42",
          "41",
          "42"
        ]
      ]
    },
    "normalize": {
      "type": "boolean",
      "default": true
    }
  }
}
Output schema
{
  "type": "object",
  "additionalProperties": true
}