Nicheloom visits: 60,840 all-time · 7,913 in the last 7 days

The opportunity tracker for new startups.

Today: 1,919 launches · Niche of the day: browser-based utility tools · See what launched today →

We fingerprinted 178 AI models' writing styles and similarity clusters

Get picks like this daily. Today's launch numbers, the top launches, AI and tech news, and a daily opportunity pick, straight to your inbox.

Want to reach these readers? Sponsor an issue.

We have a dataset of 3,095 standardized AI responses across 43 prompts. From each response, we extract a 32-dimension stylometric fingerprint (lexical richness, sentence structure, punctuation habits, formatting patterns, discourse markers).Some findings:- 9 clone clusters (>90% cosine similarity on z-normalized feature vectors) - Mistral Large 2 and Large 3 2512 score 84.8% on a composite metric combining 5 independent signals - Gemini 2.5 Flash Lite writes 78% like Claude 3 Opus. Costs 185x less - Meta has the strongest provider "house style" (37.5x distinctiveness ratio) - "Satirical fake news" is the prompt that causes the most writing convergence across all models - "Count letters" causes the most divergenceThe composite clone score combines: prompt-controlled head-to-head similarity, per-feature Pearson correlation across challenges, response length correlation, cross-prompt consistency, and aggregate cosine similarity.Tech: stylometric extraction in Node.js, z-score normalization, cosine similarity for aggregate, Pearson correlation for per-feature tracking. Analysis script is ~1400 lines.

HN · launched Apr 8, 2026 · 78 upvotes (top 12%) · 1 view

Niche: specialized ai analytics platforms · 123 launches, 12 in the last 7 days ▼ -29% · Explore this niche →

Category and tags

  • Horizontal
  • Analytics & BI
  • Developer
  • AI-native
  • Hobby / open-source project

Visit website ↗ Source listing ↗

Could you build this?

Partial Extracting stylometric features from text via standard NLP libraries is straightforward, but assembling, cleaning, and normalizing a benchmark dataset across 178 models requires substantial automated evaluation pipelines and API compute.

What it would actually take: The implementation requires automated prompt evaluation harnesses querying dozens of model endpoints via standard APIs (OpenAI, Anthropic, open-source model providers via vLLM), followed by a feature extraction pipeline computing 32 linguistic and statistical metrics (lexical diversity via TTR/Yule's K, POS distribution via spaCy, sentence entropy, and formatting tokens). Dimensionality reduction and clustering (UMAP, HDBSCAN, or hierarchical clustering) must then be applied and validated across thousands of samples.

Discussion

20 comments analyzed.

Competitors mentioned: models.dev, arena.ai, HuggingFace

Concerns raised: Methodology lacks linguistic theory and uses arbitrary metrics/thresholds, Claims (75% similarity = 'writes the same') unsubstantiated without prompts/responses shown, Accessibility issues: muted colors on dark background, poor contrast ratios, Suspicious claim that Opus and Gemini Flash share 99% style similarity, No explanation of how 32 dimensions were selected or if prompts were optimized

Competitors

Other products that read as similar to this one : 864 launches clear the similarity bar, closest 8 shown.

Attention rank: #107 of 865 (itself plus its competitors, highest first, normalized so YC and Product Hunt are compared fairly).

Launched 160 days after the earliest competitor.

The niche, labels, vibe code rating and competitor comparison on this page are analysis by Nicheloom. If you use them, please credit nicheloom.com. How you may use our data.