Fronset Benchmark

Best LLMs for Topic Sequence Optimization

Orders supplied topics for a coherent long-form presentation using logical dependencies, increasing complexity, narrative flow, and reader value. Illustrative uses include ordering an implementation plan, customer guide, software course, board narrative, research report, document

Cost mode:

Category: Topic Organization & Clustering · Typical I/O: 3788→5853 tokens

Run this task on Fronset — request an invitation

Models

Frontier on this task: Gemini 3.5 Flash at 8.98 / 10. Quality bar at 90%: 8.08.

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.

Cost breakdown

ModelQualityConfidenceCost / 1k runsOverpayMode
Tencent Hy3 ★ OpenRouter8.24 / 10 CI [7.96, 8.52]HIGH$0.78best valuebatch
Gemini 3.5 Flash best Gemini8.98 / 10 CI [8.76, 9.21]HIGH$21.5127xbatch
Moonshot Kimi K3 Moonshot AI8.48 / 10 CI [8.16, 8.81]MEDIUM$88.67113xbatch

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: 3788 input tokens → 5853 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 dependency ordering, narrative coherence, progression, local transitions, reader orientation, and use of every topic exactly once. Identifier completeness is deterministic.

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.

{
  "$defs": {
    "TopicSequenceItem": {
      "description": "Single topic in the recommended sequence.",
      "properties": {
        "topic_id": {
          "description": "The ID of the topic (exactly as provided in input)",
          "title": "Topic Id",
          "type": "integer"
        },
        "topic_name": {
          "description": "The name of the topic (exactly as provided in input, do not modify)",
          "title": "Topic Name",
          "type": "string"
        }
      },
      "required": [
        "topic_id",
        "topic_name"
      ],
      "title": "TopicSequenceItem",
      "type": "object"
    }
  },
  "description": "Output model for determining the optimal sequence of topics in a report.\nThe LLM analyzes topic content and determines the best reading order.",
  "properties": {
    "sequence_reasoning": {
      "default": "",
      "description": "Brief explanation of why this sequence was chosen (e.g., logical flow, building complexity, narrative arc).",
      "title": "Sequence Reasoning",
      "type": "string"
    },
    "sequenced_topics": {
      "description": "Ordered list of topics in the recommended reading sequence. First topic appears first in report.",
      "items": {
        "$ref": "#/$defs/TopicSequenceItem"
      },
      "title": "Sequenced Topics",
      "type": "array"
    }
  },
  "required": [
    "sequenced_topics"
  ],
  "title": "TopicSequenceOutput",
  "type": "object"
}

Prompt templates

The published template for this task.

TOPIC_SEQUENCE_SYSTEM_PROMPT + TOPIC_SEQUENCE_USER_PROMPT (300 calls in window)

System prompt

You are an expert editor specializing in organizing research reports for optimal readability and narrative flow.

Your task is to determine the best sequence for presenting a set of topics in a comprehensive research report. Consider:

1. **Logical Flow**: Topics that provide context or background should come before more specific analysis
2. **Building Complexity**: Start with foundational topics, then move to more nuanced analysis
3. **Narrative Arc**: Create a compelling story that guides readers from introduction to conclusion
4. **Dependencies**: Some topics may reference or build upon information from others
5. **Reader Engagement**: Consider what order will keep readers engaged and informed

For each topic, also provide a polished section title that may differ slightly from the raw topic name to improve report cohesion.

Output a sequence that would work well for a professional research document.

## Required Output Format
Your response MUST be a single, valid JSON object conforming to this schema:
```json
{schema_json_string}
```

User prompt

Subject: {subject_name}

I have {topic_count} topics to organize into a coherent research report. Please determine the optimal sequence for presenting these topics.

Topics to sequence:
{topics_json}

Task: Analyze these topics and determine the best order for presenting them in the final report. Consider:
- What provides necessary context for other topics?
- What logical flow would help readers understand the subject?
- How can we create a compelling narrative arc?

For each topic, provide:
1. The topic_id (exactly as shown above)
2. The topic_name (exactly as shown above)

IMPORTANT: Only return the sequence order. Do NOT modify topic names or add numbering - the system handles that automatically.

The required JSON output schema is provided in the system prompt.