What's new in
LumiChats.

A security patch that strengthens how LumiChats protects your account during the payment flow — your subscription can only ever be activated by you.

Deep Research Mode is now live — LumiChats can search the web in real time and bring back fresh, cited results directly inside your conversation. Chat Mode auto-detects when a question needs live data and searches silently. Agent Mode gives you a full research tool the AI controls step by step — best for thorough, multi-source investigations.

LumiChats Labs launches — a public research hub showcasing all six open-source fine-tuned models built by the LumiChats team. Each model page includes full architecture specs, training details, benchmarks, BibTeX citation, and a direct Hugging Face download link.

Your account is more reliably protected. Payments confirm faster. PDF study sessions communicate their limits clearly. Blog article suggestions are smarter and more editorial. LumiChats is now accurately discoverable by AI-powered search engines. Blog post titles restored with brand name for better Google recognition.

SEO overhaul across the homepage and blog. The homepage now qualifies for Google FAQ rich results. Social share image upgraded to 1200×630. User count updated to 14,000+. Pricing updated to universal 'under $1/day'. Blog articles now auto-link to all seven competitor comparison pages. All seven alternatives pages re-enabled for Google indexing after an accidental noindex was found and removed.

The landing page is now sharper and more responsive across every screen size. Five bugs fixed including a silent animation issue breaking FAQ interactions. LumiChats Offline and Quiz Hub now appear in the top announcement ticker. Blog articles gain a reading streak tracker, a helpful/not-helpful feedback widget, and a pull-quote share button that correctly dismisses on mobile scroll.

The blog opens faster and scrolls more smoothly than before. We also made several behind-the-scenes improvements to keep the site healthy, fast, and reliable for everyone.

LumiChats Offline LLM gets its own dedicated documentation hub — a fully designed /docs section launches today with an index page and a comprehensive offline setup guide. Run powerful AI privately on your Windows PC with no internet, no GPU, and no data collection.

The chat page now opens faster and with a more accurate view of your account on every visit. Five key pages across the site have sharper, more focused keyword strategies so each one surfaces in genuinely different searches. The codebase now passes a stricter quality bar with zero compile issues.

Blog posts now show a unique preview image for each article when shared on any platform. FAQ answers are now eligible to appear directly in Google search results. Fonts load faster across the site. Star ratings from verified reviews are now eligible in Google search. The blog internal linking system covers 8× more topics. Blog pages load faster, and search engines now crawl the blog more efficiently.

The chat sidebar now opens instantly the moment you land on the page, conversations load on demand, fonts render cleanly from the first visit, and switching between chats is noticeably snappier and more reliable.

Behind-the-scenes SEO upgrades that make LumiChats content more visible in Google Search — the blog now inherits author credentials on every post, all 200+ glossary terms are eligible for Google's definition rich-result box, and the blog cross-linker now covers more topics automatically.

A new Authors page introduces the people behind LumiChats — published author Aditya Kumar Jha and CTO Shikhar Burman. Blog posts now share beautifully on every platform with a proper 1200×630 image card, related article suggestions are matched by topic rather than URL keywords, and every post now cross-links to related articles inline in the server-rendered HTML.

A behind-the-scenes security update that makes your account meaningfully harder to misuse — every request to LumiChats is now verified against your actual login session, so your subscription and message quota can only ever be used by you.

Study mode gets a meaningful upgrade across the board — follow-up questions now actually work, large documents are handled safely, and a round of reliability fixes make PDF study sessions noticeably smoother and faster.

Agent Mode now picks up right where it left off after a page refresh — files from your last session are automatically restored to the workspace, stuck loading spinners from interrupted tasks clear themselves, and the agent always has an accurate picture of what's already there before it starts work.

The thinking panel no longer jumps to the wrong position during streaming, and the file viewer now renders README and markdown files with proper formatting — syntax-highlighted code, GitHub-style tables, and a per-block copy button.

A focused quality-of-life update for Agent Mode — file writing is now visibly smooth, the agent stops after finishing your task instead of creating more files on its own, and the thinking panel always appears in the right place.

Agent Mode gets a major upgrade — it now thinks between every step, you can watch files being written line by line in real time, and the terminal panel opens automatically so you never miss what the agent is doing.

The chat screen now feels more alive and polished — you'll see a gentle waiting animation the moment you send a message, conversations look cleaner with less visual clutter, and your own messages are easier to read at a glance.

Your chats now save more reliably — messages no longer show false save errors when switching between conversations, and the app is faster to respond when saving because it no longer makes an unnecessary network call every single time.

Chat experience significantly smoother — streaming now flows in a steady typewriter style instead of jerky bursts. All token and message limits removed for paid users. Multiple security patches applied across the platform.

Blog articles now open instantly — even on the very first visit after a long break. Pages that only needed a title or date no longer wait for all 315 articles to load first.

Blog articles now show image previews when shared anywhere, load faster, scroll smoother, and surface better related reading mid-article. Category badge colours fixed on 62 posts. 'Read next' links inside articles are now actually clickable.

Major SEO and social sharing update — link previews now show rich images on WhatsApp, Twitter, LinkedIn, and iMessage across all key pages.

A fun animated 404 page, broken blog links now redirect to the right article, and the glossary page looks better on all phones.

Glossary term pages fully overhauled — multi-platform sharing, mobile section navigation, visual track progress, fixed next/prev arrow layout, and better retention ordering.

Security update — blocked a new wave of aggressive bots that were hammering the site and draining server capacity away from real users. Glossary and changelog pages now have dedicated traffic limits.

Blog posts now automatically surface related glossary definitions inline — click any highlighted AI term while reading to get a full explanation without leaving your flow. Also fixed a bug where most articles were missing the Read Next section entirely.

Blog content is now shareable. Every article and blog listing card has direct share buttons for X, LinkedIn, and WhatsApp — plus a save-for-later bookmark that persists across sessions.

Agent Mode boot fixed — CSP was silently blocking WebContainer. Reasoning now streams in real time. Terminal output streams smoothly instead of dumping all at once.

Google OAuth fully restored, authentication hardened with PKCE session exchange, and Content Security Policy tightened across all non-auth routes.

Full security audit — XSS hardening, CSP, database RLS, memory fixes, and data quality improvements across the entire codebase.

Agent Mode stability patch — infinite loop fix, WebContainer boot fix, step counter wired, and npm install fix for Word/Excel file creation.

New premium models, study content expansion, and security updates.

Rich HTML rendering in chat — cards, charts, math, SVGs, and code layouts rendered natively.

Voice input, shareable chat links, and AI glossary.

SEO expansion — alternatives, features, use-cases, and footer navigation added.

Complete UI overhaul — every surface redesigned for a cleaner, faster experience.

User tools panel, persistent Study Mode sessions, and quiz history tracking.

Agent Mode launched — AI that writes and executes real code in your browser.

Study Mode, Quiz Hub, and note generation added.

RAG pipeline, vector embeddings, and security hardening.

LumiChats public launch — multi-model AI chat with pay-per-day pricing.