{
  "title": "Output token limit",
  "description": "Identify the maximum response or generated-artifact budget available through the exact harness.",
  "slug": "output-token-limit",
  "locale": "en",
  "seoTitle": "Output token limit comparison — Can My Agent Use",
  "socialTitle": "Output token limit",
  "socialDescription": "Compare maximum response budgets and harness-specific output caps.",
  "llmSummary": "Output-token limits are recorded separately from total context because harnesses may reserve, cap, or truncate generated text and artifacts independently.",
  "audience": "Engineers sizing generated code, reports, and structured output.",
  "contentKind": "feature",
  "status": "published",
  "tags": [
    "context",
    "output",
    "tokens",
    "limits"
  ],
  "updated": "2026-08-28T00:00:00.000Z",
  "published": "2026-08-28T00:00:00.000Z",
  "category": "models-context",
  "summary": "Record the maximum generated response budget by exact model and harness mode.",
  "specLabel": "Measured product property",
  "aliases": [
    "max output tokens",
    "response length limit",
    "generation limit"
  ],
  "capabilityKind": "atomic",
  "parent": "models-and-context",
  "related": [
    "context-window",
    "structured-output"
  ],
  "relations": [],
  "highlight": false,
  "notes": [],
  "issues": [],
  "resources": [
    {
      "title": "Methodology",
      "href": "/methodology",
      "kind": "note"
    },
    {
      "title": "OpenAI — Create a model response",
      "href": "https://developers.openai.com/api/reference/cli/resources/responses/methods/create",
      "kind": "docs",
      "publisher": "OpenAI",
      "reviewedAt": "2026-08-28"
    }
  ],
  "support": [],
  "html": "/features/output-token-limit",
  "markdown": "/features/output-token-limit.md",
  "json": "/api/v1/features/output-token-limit.json",
  "body": "This row tracks the maximum response or artifact generation budget exposed by the harness. The number may be lower than a model API limit and may vary for chat, code edits, tool calls, structured output, or background tasks.\n\nEvidence should identify units, model and mode, defaults, user controls, stop behavior, and whether the harness continues long results across messages or files. UI clipping, transport truncation, and model generation limits should not be collapsed into one cause."
}