Cost mode:

Sustained compositional skill, voice consistency, coherent extended prose.

6 capabilities in this category.

Task-by-task breakdown

Substack Newsletter

Pooled TT for Substack opener and summary newsletter generation. Same role/voice; opener is shorter announcement, summary is longer recap with per-article links.

ModelQuality (% of best)ConfidenceOverpay
NVIDIA Nemotron-3 Nano 30B-A3B 93%RANKEDbest value
MiniMax M399%RANKED2.9x
Tencent Hy392%HIGH4x
DeepSeek V4 Pro99%HIGH4.8x
GPT-5.6 Luna91%RANKED5.1x
GPT-5.6 Terra90%HIGH9.4x
Claude Haiku 4.598%RANKED12x
Qwen 3.6 Plus97%HIGH15x
Qwen 3.7 Plus96%RANKED16x
NVIDIA Nemotron-3 Ultra 550B98%RANKED17x
Claude Sonnet 5 best100%RANKED19x
Qwen 3.6 Flash97%RANKED19x
Gemini 3.1 Pro Preview96%HIGH19x
Gemini 3.5 Flash99%RANKED24x
Kimi K2.699%RANKED26x
GPT-5.6 Sol95%RANKED30x
Grok 4.599%RANKED36x
Meta Muse Spark 1.195%RANKED46x
GPT-5.597%HIGH75x

Task detail →

Landing Page Section Generation

Designs up to 10 thematic sections for a publication's Ghost CMS landing page. Mixes 1-2 public sections (introductory / overview) and the remainder premium (in-depth analysis). Section names are 2-5 …

ModelQuality (% of best)ConfidenceOverpay
Tencent Hy3 92%HIGHbest value
MiniMax M397%MEDIUM1.7x
Qwen 3.7 Plus93%RANKED1.8x
GPT-5.6 Luna98%MEDIUM2x
GPT-5.6 Terra best100%MEDIUM3.8x
NVIDIA Nemotron-3 Ultra 550B94%RANKED4.2x
Qwen 3.6 Flash90%RANKED5.9x
Meta Muse Spark 1.193%MEDIUM9.3x
Grok 4.595%RANKED9.8x

Task detail →

Onboarding Chapter Outline Generation

Designs the chapter outline for a new analysis template — 5-10 chapters typically, each with a snake_case code and a detailed user_requirement specifying what to cover, what questions to answer, and …

ModelQuality (% of best)ConfidenceOverpay
NVIDIA Nemotron-3 Super 120B 93%RANKEDbest value
Tencent Hy391%MEDIUM1.6x
Qwen 3.7 Plus best100%RANKED2.9x
Qwen 3.6 Flash98%RANKED3.3x
GPT-5.6 Luna95%RANKED4.6x
NVIDIA Nemotron-3 Ultra 550B99%RANKED6.9x
Claude Sonnet 594%RANKED9.7x
GPT-5.6 Terra97%RANKED11x
Meta Muse Spark 1.191%HIGH12x
Grok 4.598%RANKED13x
GPT-5.6 Sol99%RANKED29x

Task detail →

Author Voice Generation

Crafts a 'soul' document — a detailed personality and voice spec for an AI author persona named after a deceased historical figure. Channels the real figure's intellectual style, values, and …

ModelQuality (% of best)ConfidenceOverpay
NVIDIA Nemotron-3 Super 120B 90%HIGHbest value
DeepSeek V4 Flash95%RANKED1.8x
MiniMax M3 best100%RANKED2.1x
GPT-5.4 Nano90%RANKED4.6x
GPT-5.6 Luna93%RANKED4.7x
NVIDIA Nemotron-3 Ultra 550B92%MEDIUM6.7x
DeepSeek V4 Pro95%RANKED7.6x
Qwen 3.7 Plus94%RANKED8.2x
Claude Sonnet 594%RANKED12x
GPT-5.6 Terra92%RANKED12x
Gemini 3.5 Flash98%RANKED13x
Meta Muse Spark 1.195%RANKED15x
Grok 4.596%RANKED16x
Claude Haiku 4.592%RANKED19x
GPT-5.6 Sol94%RANKED25x
Claude Sonnet 4.694%RANKED58x
GPT-5.595%MEDIUM117x

Task detail →

Confidence — how sure we are about the quality score (more judgments + more agreement = higher confidence): RANKED many independent judges scored this model's outputs and their agreement is very high (most confident) — HIGH many judges have scored it and they mostly agree (well-pinned) — MEDIUM enough judges have weighed in to publish, but they disagree more than we'd like (treat with a small grain of salt). LOW-confidence cells are hidden everywhere on the site. See the methodology for the exact thresholds.