Changelog

You can now compare up to 20 AI models in a single query — up from 10. Mega and Admin tier users get the full 20-model limit. Run your most complex prompts against the entire model lineup at once.

Two new pages are live: /support features a full FAQ covering Getting Started, Billing, Models, and Privacy — plus a contact form that delivers messages directly to the team. /docs is a complete how-to guide with slash command reference (/rank, /estimate, /compare), model comparison table, and tips & tricks.

This very page! A public-facing changelog at /changelog shows all updates in a clean timeline with version badges, color-coded category pills, and grouped-by-month layout. Admins can create, publish, and manage entries from Admin → Changelog.

Admins can now send targeted notifications to individual users or entire subscription tiers (Trial / Pro / Mega) from Admin → Notifications. Features user search autocomplete, notification type picker, live preview, and a broadcast history table showing recipient counts.

Full SEO overhaul: per-page dynamic meta tags, Open Graph and Twitter Card support on every page, Schema.org JSON-LD (WebApplication + Organization), llms.txt for AI crawlers, robots.txt, and sitemap.xml. Shared result pages now include a rich marketing preview card so every link shared on X or LinkedIn pulls a compelling preview.

Full Terms of Service and Privacy Policy are now live at /terms and /privacy. Covers data collection, anonymous research use of prompt replies, GDPR/CCPA rights, security practices, and contact information. Linked from the site footer.

When AI models significantly disagree on your prompt, a disagreement alert banner now includes a one-click Share on X button. The tweet is pre-filled with the spread size, severity emoji, a prompt snippet, and a shareable result URL — generated automatically the moment a disagreement is detected.

A bell icon in the navigation bar shows your unread notification count (polls every 30 seconds). The dropdown lists all notifications with per-type icons, relative timestamps, per-item dismiss, and mark-all-read. Admins can broadcast to all users or specific users from the Admin panel.

Compare responses from multiple AI models simultaneously with a single prompt. Features include slash commands (/rank, /estimate), prompt quality analyzer, sycophancy detection, shareable result links, query history, subscription tiers, and an admin dashboard.