Launched this week

BenchmarkAI Research
4 AI models' stock research, in your assistant
2 followers
4 AI models' stock research, in your assistant
2 followers
Stock research from ChatGPT, Claude, Gemini and Grok, inside your assistant. Read what each model's reports say about a US company, with the case for and against and dated sources. Check financials against SEC filings, review holdings you provide, or build a portfolio using the BenchmarkAI method, reviewing and approving each step. Connect through MCP to Claude, ChatGPT, Claude Code, Codex or other remote MCP hosts. Free for personal use. In beta.








Free Options
Launch Team

Render Sandboxes: a computer for every agentIsolated compute, persistent state and a real filesystem
Promoted
I'm Mike, founder of BenchmarkAI. We built BenchmarkAI Research so you can use the stock research ChatGPT, Claude, Gemini and Grok wrote inside the assistant you already use: read what each model said about a company, check its financials, review your holdings, or build a portfolio using the BenchmarkAI method.
On December 1, 2025, the first 12 portfolios started, with the 4 models doing the research and the portfolio decisions. 160 index portfolios followed on October 5, 2026. In the October 30 rebalance alone, the models wrote more than 18,000 equity research reports on about 1,300 companies. I run the system and provide no investing judgment.
Each report keeps the date it was written, so you can read what a model said before the results came in. Each answer links to the reports behind it, and where evidence is missing, the answer says so. The current versions are GPT-6 Astra, Claude Fable 5.1, Gemini 4 Argon and Grok 4.7; earlier reports came from earlier versions.
On November 18 I published how the system works: https://x.com/mgonzalez/status/2123100176420053372
It's an MCP server: add https://www.benchmarkai.app/api/mcp to Claude, ChatGPT, Claude Code, Codex or any host that supports remote MCP, sign in, and ask. Steps for each host: https://www.benchmarkai.app/agents/plugin/docs
3 things to try:
A company. "I'm reviewing Microsoft. What supports the investment case, and what could undermine it?" You get how each model's reports rated it, the case for and against, and links to the reports, with dates.
Your holdings. "Review these holdings against BenchmarkAI's latest research: AAPL 50%, MSFT 40%, 10% cash." Before anything is sent, your assistant shows the date, tickers, weights and cash.
A portfolio using the BenchmarkAI method. "Help me build a portfolio of durable cash-generating businesses." Your assistant screens candidates from the models' research, reads the reports behind them and drafts a hypothetical portfolio, stopping for you to review and approve each step. The video shows this workflow on sample data.
The portfolios hold no real money. BenchmarkAI publishes research and indexes; nothing in this product is investment advice. Nothing is bought, sold or managed, and no brokerage account is connected.
It's free for personal use; using it at work needs a Work seat. It's in beta. Try a company you know well: where is the evidence useful, and what is missing? Tell me what breaks.