You can now keep using Driven for free without a paid subscription. The Free plan includes Auto and select light-tier models for everyday, lightweight investment research.
We put 11 leading AI models through 15 categories of real-world investment-agent tasks. The results looked very different from general-purpose benchmarks
Consider a common portfolio-risk request. A user provides one month of fund NAV data and only a few months of monthly closing prices for the portfolio’s U.S. equity holdings. One model calculates a 63.3% annualized return and a Sharpe ratio of 6.0, then adds a disclaimer. Another correctly identifies that the data is insufficient
The better response lies between those extremes: calculate what the available data can support, explain what cannot be determined, and give the user a useful partial result without pretending the missing information exists.
Driven Connectors are now available to everyone. You can connect Driven to seven brokerages including IBKR, Moomoo, Longbridge etc., as well as a growing range of financial data services and productivity MCP tools. More useful MCP integrations are on the way.
Driven now supports Gemini 3.7 Flash across text, images, video, audio, and PDFs. It is our recommended model in its tier for quickly understanding and organizing multimodal research materials.
More Accurate Hong Kong Order Checks
Hong Kong market data now includes board lot sizes. The Agent can validate order quantities against actual trading units before submission, reducing errors caused by board lot mismatches.
Citations appear in real time: Sources now appear as the Agent generates its response, with key claims underlined as they are written. You can read and verify the evidence without waiting for the full response to finish.
Clearer citation cards: Hover over a citation to see the exact supporting passage highlighted in the original content, along with a direct link to the source page. Tracing and verifying information is now faster and more straightforward.
Driven now integrates Claude Opus 5, available directly from the model list. It supports up to one million tokens of context, along with image and PDF inputs.
Skill Store for everyone, Gemini 3.6 Flash, and Hong Kong options
Shiny New ThingsEven Better NowWe Fixed It
Skill Store
Everyone can now discover and install Skills from the Skill Store. Each Skill packages the instructions, tools, and workflow needed for a specific job, helping your Agent handle research, writing, and file tasks more consistently. Find the Skill you need, install it, and use it directly in a conversation. You can also publish your own Skills to the Store and keep them updated over time.
Gemini 3.6 Flash
Driven now supports Gemini 3.6 Flash. It accepts multimodal inputs including images, video, audio, PDFs, and text files, making it a strong choice for tasks that need speed, reasoning, and source understanding.
Driven Agent can now query real-time, complete options chains for US stocks and ETFs. Filter by calls or puts, expiration date, and strike range, then inspect quote timestamps, volume, open interest, implied volatility, Greeks, and bid-ask data. Complex options strategies are now faster to research and compare.
More reliable built-in Skills
Official Skills including Stock Screener, Portfolio Monitor, Market Pulse, Insider Tracker, Sector Radar, and Data Lab are now more stable and reliable. Across stock screening, portfolio monitoring, sector tracking, and data analysis, common research workflows now run more consistently and produce more dependable results.
Six new technical indicators: You can now calculate SMA, EMA, RSI, MACD, Bollinger Bands, and KDJ directly for supported US stocks, Hong Kong stocks, and A-shares.
US pre-market and after-hours data: Extended-hours price queries now return the latest traded price, trade time, and volume.
More accurate analysis for stocks, ETFs, and funds: Returns, profit and loss, drawdowns, and technical indicators now use adjusted price data by default. Driven also identifies the security type before selecting an analytical method, preventing company-specific approaches such as DCF from being incorrectly applied to ETFs or funds.
Driven now supports the full GPT-5.6 model lineup across three tiers: Sol, Terra, and Luna.
You can choose the model that best fits the complexity of your task. Long context and reasoning usage are also calculated more accurately for each tier.
File preview got the biggest overhaul: you can open multiple files at once and switch between them with tabs, collapse or expand the preview panel, and it remembers per conversation whether you had it open, restoring that state when you return
The Accounts page was improved too, with tabs for switching between accounts, and the account panel expands automatically when you act on an account mid-conversation.
When an Agent cites data, news, or a file in its answer, you'll now see it underlined. Hover or tap on it and you'll get the full picture: where the information came from, when it was last updated, which tool pulled it, and the exact snippet behind it. So instead of wondering where a number came from, you can just check for yourself. It's a big step up in how transparent and trustworthy each answer feels.
Financial data for US-listed Chinese stocks, Hong Kong stocks, and A-shares got a full upgrade. All three major statements now pull from sources aligned with local accounting standards, with better coverage and significantly faster refresh cycles. Data routing rules were also refined so the Agent fetches the right data more accurately.
Three new data capabilities specific to A-shares: capital flow tracking, shareholder structure, and company intelligence: thematic linkages, revenue segment commentary, share pledging, valuation bands, historical financials, peer comparison, and more.
Instead of manually topping up Credits and deducting per use, your usage now resets automatically on two rolling windows: every 5 hours and every 7 days. And your total allowance? Significantly higher than before.
Driven Skills. Institutional-grade. One prompt away.
Driven
· 8 mins read
Ask a generic AI assistant to "analyze this stock" and you'll get something that sounds like a finance textbook wrote a tweet: confident, readable, and quietly wrong in the places that matter. The model blends valuation methods it doesn't name, leans on whatever it saw most often during training, and happily fabricates a price target with a citation that doesn't exist. For a curious investor, that's a party trick. For anyone with real money on the line, it's a liability.
This is the gap we built the Driven Skills Market to close.
A Skill in Driven is not a prompt template. It's a repeatable, named workflow grounded in a specific investment framework
Audio & video understanding, plus a refreshed Schedules
Shiny New ThingsEven Better NowWe Fixed It
Schedules, redesigned
The Schedules panel is now split into "Latest" and "Manage" views, so notifications and long-running tasks don't step on each other. Unread badges and model icons make it easy to scan at a glance. If you've bound Telegram, scheduled task notifications push directly to the right channel
Driven on Telegram. Plus Skills you can build yourself
Shiny New ThingsEven Better NowWe Fixed It
Driven on Telegram
Driven is now on Telegram. Ask your Agent questions or kick off deep research right in Telegram. Schedule results push back to you automatically. Add your Bot to a Telegram Group and your team can @-mention it to ask questions or issue commands (you can manage who's authorized from the web). Now Driven goes wherever you go.
Skills are live—your investment research toolkit, built in
Shiny New ThingsEven Better NowWe Fixed It
Skills library
Skills are here. We've built in 8 investment research skills that cover the full workflow, from macro to individual stocks. Your Agent picks the right Skill automatically based on your question, and you can toggle them anytime in the right panel.
Chat got a full redesign. The new three-column layout makes creating conversations, switching portfolios, and managing schedules feel natural. You can pin and rename chats, and they auto-sort by recent activity. Titles are generated from the conversation itself, and any chat running a live schedule floats to the top