Basic, Standard and Deep Debate reports. Up to 25 AI agents (21 specialist analysts plus the debate and committee team) build the bull case, the bear case and the risks, then show their work.

Basic reports run on every plan, Standard from Starter and Deep Debate from Pro. The full 25-agent roster runs in Pro+ Deep Debate. Every report ends in one research view with a confidence score, the evidence behind it and what would change the thesis.

Confidence now measures research quality times depth, and is kept separate from conviction and from how much the agents agree.

Each report shows category views, specialist-fed scenario ranges and the top cards that drove the view, so you can see why the research lands where it does.

Independent AI model families (OpenAI, Claude, Gemini, Grok and DeepSeek) each read the same evidence and state their own view, category by category.

See where the models agree and where they split. Agreement is not a measure of being right: the models are not fully independent. On Standard and Deep Debate reports, Starter and up.

Reports list the assumptions their thesis depends on, in a checkable form where they can (a metric, a price level, a date), and what would change the view.

Each assumption is re-checked as new data arrives and marked on track, at risk or broken. Alert notifications follow after a calibration period.

See which evidence each side of the research rests on, where both sides cite the same fact, and which claims are thin.

The map is built from what the report recorded, so the page and the PDF, Word and CSV exports say exactly the same thing.

Run a stock again and compare the new thesis with the old one: what changed in the view, the confidence and the assumptions.

For You, Following, Trending and Latest. Follow stocks and researchers, scan ticker stacks and freshness chips on every card, and hide or mute what you do not want to see.

Full access from Starter. Free accounts get a 7-day trial from their first visit.

Share a Standard or Deep Debate report to the Feed with a GIF. Readers reply with a GIF or react with thumbs up, thumbs down, mind-blown or thinking. GIFs are the replies.

Reactions rate the research, never the stock. Share images carry the company logo and the AI model logos, and anyone who disagrees can run their own view.

Everyone publishes under a generated research name and avatar; verified researchers carry a check mark.

Published reports get their own public research page, and a challenge lets anyone run their own view of the same stock and compare the two.

The Most Promising for Research carousel is refreshed every day with Valarn AI picks, next to insider and Congress trades, the market map, the catalyst calendar and Valarn Media news.

Picks are ideas worth researching, not recommendations. The daily Market Brief keeps you oriented before the open.

P/E against its own 5-year history, an SEC filings panel, backtesting, and options intelligence on Pro+.

Backtests are educational historical simulations, never a prediction. Options intelligence is included on Pro+, Team and Enterprise.

One full Pro+ report on a real stock, free for everyone, every week.

Ask Valarn AI about a report and get answers grounded in its evidence, listen to an audio brief or podcast, and export to PDF, Word or CSV.

Every report and export now states the Valarn version that produced it.

Credit pricing that scales with report depth, top-up credit packs on paid plans, and Give 3 Get 3 referrals.

Valarn in English, Spanish, French, German, Portuguese, Japanese, Chinese, Arabic and Turkish, plus a status page that shows how every service is running.