Cost mode:

Category: Content Summarization & Synthesis · Rail: absolute · Typical I/O: 41168→2659 tokens

Models

Frontier on this task: GPT-5.5 at 7.93 / 10. Quality bar at 90%: 7.14.

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 clears the bar but whose CI low does not.

ModelQuality scoreCI lowCost / 1k runsvs best value
MiniMax M37.50 / 107.34$5.79best value
GPT-5.4 Nano7.47 / 107.21$14.652.5x more expensive
Qwen 3.6 Flash7.36 / 107.08$17.673.1x more expensive
Qwen 3.7 Plus7.84 / 107.65$20.263.5x more expensive
GPT-5.6 Luna7.66 / 107.35$20.773.6x more expensive
Gemini 3.5 Flash7.57 / 107.18$32.215.6x more expensive
DeepSeek V4 Pro7.89 / 107.48$33.835.8x more expensive
Grok 4.57.92 / 107.81$39.146.8x more expensive
Claude Sonnet 57.60 / 107.50$42.957.4x more expensive
Claude Haiku 4.57.67 / 107.37$49.608.6x more expensive
GPT-5.6 Terra7.73 / 107.44$51.598.9x more expensive
Meta Muse Spark 1.17.55 / 107.12$63.3411x more expensive
Kimi K2.67.67 / 107.26$127.8722x more expensive
Claude Sonnet 4.67.81 / 107.46$172.5730x more expensive
GPT-5.57.93 / 107.47$321.5456x more expensive
Qwen 3.5 Flash6.83 / 106.54$10.231.8x more expensive
Qwen 3.6 Plus7.02 / 106.65$64.1111x more expensive
GPT-5.4 Mini6.95 / 106.57$22.733.9x more expensive

Cost breakdown

ModelQualityConfidenceCost / 1k runsOverpayMode
MiniMax M3 MiniMax7.50 / 10 CI [7.34, 7.66]RANKED$5.79best valuebatch
GPT-5.4 Nano OpenAI7.47 / 10 CI [7.21, 7.73]HIGH$14.652.5xbatch
Qwen 3.6 Flash Alibaba Cloud (DashScope)7.36 / 10 CI [7.08, 7.64]HIGH$17.673.1xbatch
Qwen 3.7 Plus Alibaba Cloud (DashScope)7.84 / 10 CI [7.65, 8.03]RANKED$20.263.5xbatch
GPT-5.6 Luna OpenAI7.66 / 10 CI [7.35, 7.97]MEDIUM$20.773.6xbatch
Gemini 3.5 Flash Gemini7.57 / 10 CI [7.18, 7.96]MEDIUM$32.215.6xbatch
DeepSeek V4 Pro DeepSeek7.89 / 10 CI [7.48, 8.30]MEDIUM$33.835.8xbatch
Grok 4.5 xAI7.92 / 10 CI [7.81, 8.03]RANKED$39.146.8xbatch
Claude Sonnet 5 Anthropic7.60 / 10 CI [7.50, 7.70]RANKED$42.957.4xbatch
Claude Haiku 4.5 Anthropic7.67 / 10 CI [7.37, 7.97]HIGH$49.608.6xbatch
GPT-5.6 Terra OpenAI7.73 / 10 CI [7.44, 8.01]HIGH$51.598.9xbatch
Meta Muse Spark 1.1 Meta7.55 / 10 CI [7.12, 7.99]MEDIUM$63.3411xbatch
Kimi K2.6 Moonshot AI7.67 / 10 CI [7.26, 8.08]MEDIUM$127.8722xbatch
Claude Sonnet 4.6 Anthropic7.81 / 10 CI [7.46, 8.16]MEDIUM$172.5730xbatch
GPT-5.5 best OpenAI7.93 / 10 CI [7.47, 8.39]MEDIUM$321.5456xbatch

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: 41168 input tokens → 2659 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.

Prompt templates

The system + user template pair used for this task.

EXECUTIVE_SUMMARY_SYSTEM_PROMPT + EXECUTIVE_SUMMARY_USER_PROMPT (1495 calls in window)

System prompt

You are an expert financial analyst specializing in creating executive summaries for research reports.

Your task is to write a concise, compelling executive summary that:
- Highlights the most critical insights from the full report
- Identifies key trends, risks, and opportunities
- Provides actionable recommendations for decision-makers
- Uses professional, analytical tone
- Matches the target length specified in the user prompt (typically 5-10% of the full report)

Focus on the big picture. Avoid detailed data points - those belong in the chapters below.

The length target will be provided as approximate word count. Adjust the number of paragraphs as needed to match the target length while maintaining coherent structure.

## 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} ({subject_code})

Report Requirements:
{report_requirements}

Consolidated Chapters (approximately {chapters_word_count} words):
{consolidated_chapters}

Task: Write an executive summary that synthesizes the key insights from all chapters above.

Target Length: Approximately {target_word_count} words ({target_percentage}% of the full report).

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