Apply stop sequences API

Applies stop sequences to raw/streamed model output: finds the earliest stop string and returns the text cut at (and excluding) it, plus which sequence matched and where — the deterministic equivalent of an API stop parameter for post-hoc trimming. Answers 'how do I cut model output at a stop token?', 'where does the first stop sequence appear?'.

Price$0.02per request

MethodPOST

Route/v1/llm/stop-sequence

StatusLive

MIME typeapplication/json

Rate limit120/minute

Cache0s public

llmstop-sequencetruncatestreamingoutputpost-processagenttrim

API URLhttps://x402.hexl.dev/v1/llm/stop-sequence

Integration docs

Example request

{
  "text": "The answer is 42.\nObservation: done",
  "stop": [
    "\nObservation:",
    "END"
  ]
}

Example response

{
  "truncated": "The answer is 42.",
  "matched": "\nObservation:",
  "matchIndex": 17,
  "wasTruncated": true
}

Input schema

{
  "type": "object",
  "required": [
    "text",
    "stop"
  ],
  "properties": {
    "text": {
      "type": "string",
      "examples": [
        "The answer is 42.\nObservation: done"
      ]
    },
    "stop": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "examples": [
        [
          "\nObservation:",
          "END"
        ]
      ]
    }
  }
}

Output schema

{
  "type": "object",
  "additionalProperties": true
}