Best LLMs for Report Outline Generation

Category: Long-form Content Generation · Typical I/O: 1171→2773 tokens

Runs on model group: All models. Inputs may reach any model in the group, not only the ranked models below.

Run this task on Fronset — request an invitation

Models

Frontier on this task: Moonshot Kimi K3 at 9.65 / 10. Quality bar at 90%: 8.68.

point-estimate floor (CI low) · upper CI (less certain) · Bars sorted by blended cost; best-value model first. Greyed rows are MEDIUM+ models whose point estimate does not clear the bar.

Model	Quality score	CI low	Cost / 1k runs	vs best value
Thinking Machines Inkling Small	9.46 / 10	9.34	$6.76	best value
Moonshot Kimi K3	9.65 / 10	9.56	$83.70	12x more expensive
Gemini 3.5 Flash Lite	8.41 / 10	8.20	$1.86	72% cheaper

Cost breakdown

Model	Quality	Confidence	Cost / 1k runs	Overpay	Mode
Thinking Machines Inkling Small ★ OpenRouter	9.46 / 10 CI [9.34, 9.57]	RANKED	$6.76	best value	batch
Moonshot Kimi K3 best Moonshot AI	9.65 / 10 CI [9.56, 9.74]	RANKED	$83.70	12x	batch

Overpay shows how much more you pay than the best-value model that clears the quality bar (marked ★) — the best-value good-enough option. "16x" means you overpay 16× — 16× that reference for no quality benefit above the bar. Typical call shape for this task: 1171 input tokens → 2773 output tokens, EMA-tracked from production traffic. Cost is the observed, all-in $ per 1,000 task runs: each model's own measured usage on this task — output verbosity, thinking/reasoning tokens, cache reads and writes, and the spend on its billed failures — priced at current list rates and adjusted by the billing overhead we actually reconcile against provider invoices. Models that answer tersely cost what they actually cost; models that think at length pay for it. Not comparable to providers' advertised $/1M list rates — this is what running the task costs, not a per-token price.

Evaluation rubric

Judge requirement coverage, section distinctness, logical sequence, useful granularity, subject relevance, evidence orientation, and quality of section requirements. Penalize boilerplate outlines and overlapping sections.

Output schema

Every answer on this task is checked against this JSON Schema, whichever model wrote it. An answer that doesn't fit counts as a model failure, and the call is retried on another model.

{
  "$defs": {
    "ChapterDefinition": {
      "description": "Definition of a single report chapter.",
      "properties": {
        "chapter_code": {
          "description": "Short code for the chapter (snake_case)",
          "title": "Chapter Code",
          "type": "string"
        },
        "chapter_number": {
          "description": "Chapter order number",
          "title": "Chapter Number",
          "type": "integer"
        },
        "chapter_title": {
          "description": "Human-readable chapter title",
          "title": "Chapter Title",
          "type": "string"
        },
        "priority": {
          "default": 100,
          "description": "Chapter priority",
          "title": "Priority",
          "type": "integer"
        },
        "relevance_threshold": {
          "default": 0.6,
          "description": "Minimum relevance score",
          "title": "Relevance Threshold",
          "type": "number"
        },
        "user_requirement": {
          "description": "What this chapter should cover",
          "title": "User Requirement",
          "type": "string"
        }
      },
      "required": [
        "chapter_number",
        "chapter_code",
        "chapter_title",
        "user_requirement"
      ],
      "title": "ChapterDefinition",
      "type": "object"
    }
  },
  "description": "Structured output for report structure generation.",
  "properties": {
    "chapters": {
      "description": "List of chapters",
      "items": {
        "$ref": "#/$defs/ChapterDefinition"
      },
      "title": "Chapters",
      "type": "array"
    },
    "description": {
      "description": "Report structure description",
      "title": "Description",
      "type": "string"
    },
    "name": {
      "description": "Report structure name",
      "title": "Name",
      "type": "string"
    }
  },
  "required": [
    "name",
    "description",
    "chapters"
  ],
  "title": "ReportStructureOutput",
  "type": "object"
}

Prompt templates

The published template for this task.

default · LLMB_REPORT_OUTLINE_GENERATION_SYSTEM + LLMB_REPORT_OUTLINE_GENERATION_USER (0 calls in window)

System prompt

Build the smallest complete section set that satisfies the stated analysis requirement for the subject, audience, focus areas, industry, and regions. Make sections mutually distinct, collectively sufficient, and logically ordered. Each section requirement must specify questions, evidence, boundaries, and expected analytical value without embedding conclusions. Use stable snake_case codes. Treat empty optional values as absent and return only the requested result. Your response must conform exactly to this output schema: {schema_json_string}.

User prompt

Inputs — subject_name: {subject_name}; subject_type: {subject_type}; industry: {industry}; focus_areas: {focus_areas}; regions: {regions}; requirement_description: {requirement_description}. Use only these inputs to complete the task defined by the system prompt.