Best LLMs for Publication Section Taxonomy Generation

Category: Long-form Content Generation · Typical I/O: 1003→907 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 9.33 / 10. Quality bar at 90%: 8.40.

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	8.97 / 10	8.85	$12.13	best value
Moonshot Kimi K3	9.33 / 10	9.27	$25.55	2.1x more expensive
Gemini 3.5 Flash Lite	8.11 / 10	7.91	$0.62	95% cheaper

Cost breakdown

Model	Quality	Confidence	Cost / 1k runs	Overpay	Mode
Thinking Machines Inkling ★ OpenRouter	8.97 / 10 CI [8.85, 9.08]	RANKED	$12.13	best value	batch
Moonshot Kimi K3 best Moonshot AI	9.33 / 10 CI [9.27, 9.39]	RANKED	$25.55	2.1x	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: 1003 input tokens → 907 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 taxonomy coverage, section distinctness, audience clarity, durable granularity, naming quality, access-model fit, and compliance with exclusions. Penalize overlapping synonyms and one-off topic labels.

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.

{
  "$defs": {
    "SectionDefinition": {
      "description": "A single section definition generated by the LLM.",
      "properties": {
        "description": {
          "description": "Brief description of what posts in this section cover",
          "title": "Description",
          "type": "string"
        },
        "name": {
          "description": "Section name (e.g., 'Brand Strategy \u0026 Positioning')",
          "title": "Name",
          "type": "string"
        },
        "suggested_visibility": {
          "default": "public",
          "description": "'public' for freely accessible content, 'premium' for subscriber-only",
          "title": "Suggested Visibility",
          "type": "string"
        }
      },
      "required": [
        "name",
        "description"
      ],
      "title": "SectionDefinition",
      "type": "object"
    }
  },
  "description": "Output model for LLM section generation.",
  "properties": {
    "sections": {
      "description": "Up to 10 thematic sections for organizing published content",
      "items": {
        "$ref": "#/$defs/SectionDefinition"
      },
      "title": "Sections",
      "type": "array"
    }
  },
  "required": [
    "sections"
  ],
  "title": "SectionGenerationOutput",
  "type": "object"
}

Prompt templates

The published template for this task.

default · LLMB_PUBLICATION_SECTION_TAXONOMY_GENERATION_SYSTEM + LLMB_PUBLICATION_SECTION_TAXONOMY_GENERATION_USER (0 calls in window)

System prompt

Propose durable navigation sections that are recognizable to the intended audience and broad enough to accumulate content without overlapping excessively. Cover the collection’s principal themes, avoid transient campaign labels and excluded names, and distinguish any public-versus-premium allocation only when the input requests it. Return stable identifiers and clear one-sentence definitions. 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 — description: {description}; subject_name: {subject_name}; excluded_section_names_block: {excluded_section_names_block}. Use only these inputs to complete the task defined by the system prompt.