Walnut vs Alpaca: Which Is Better in 2026?

Prices checked October 2026. A price marked (Oct 2026) was read on the vendor's own pricing page; others are described without a number because we could not verify one.

The AI that knows your portfolio. Ask anything in plain English, research any stock or fund, and get an honest second opinion, on the broker you already use. Read-only by default, you approve every trade, and it never holds your money. Walnut is not an investment adviser.

How it works: You connect the brokerage you already use, read-only by default, so your login stays with your broker and your money never moves. Walnut gives each holding a keep, trim or exit read from its one-year price return against the S&P 500, its momentum and its size, and flags any position above a quarter of the portfolio. You ask follow-up questions in Walnut's own assistant, or inside Claude or ChatGPT through its connector, which only reads. You can turn a thesis into a basket with target weights, and when you decide to act Walnut drafts the trades and places the ones you approve at Public plus eight other brokers including E*Trade, tastytrade and Webull.

In practice, Walnut’s AI reads your real holdings, rates each one against the S&P 500, builds baskets and drafts trades. It is an AI investing app that connects the brokerage you already use, best suited to a keep, trim or exit read on the stocks you already own, and placing the trades you approve. On connecting an account: yes (the brokerage you already use, read-only by default). On execution: places the trades you approve at Public and eight other brokers. Cost: free; Walnut Unlimited $9.99/mo on the web; no fee on assets.

One honest limitation: It has no cost basis or tax lots, so returns are one-year price returns rather than profit and loss, and it is read-only at Schwab, Fidelity and Vanguard.

A developer-first brokerage with an official MCP server for real-time data, paper trading, and execution by AI agents. Best for people who code.

How it works: Alpaca is a brokerage you talk to through code: a REST and websocket API exposes market data, paper trading, and live commission-free execution, and its official MCP server lets an AI agent pull data, test in a paper account, and place real orders programmatically. You build (or wire up an agent against) that API rather than tapping buttons in an app.

In practice, Alpaca’s AI official MCP server (data + execution). It is a brokerage API for people building their own AI agent, best suited to developers building AI agents. On connecting an account: alpaca accounts. On execution: read + trade + paper. Cost: free (open source).

One honest limitation: Developer-first; requires code and self-hosting.

The honest framing: Alpaca is about developer-level, programmatic control for building your own AI agent, while Walnut is about a no-code experience across most US brokers. Alpaca is built for developers who write code, whereas Walnut is a no-code app that sits on top of your broker.

Every tool gives something up for what it does well. Here is the honest give-and-take on each, so you can weigh the specific strengths against the limitations that come with them rather than judging on the headline category alone.

For AI investing tools, the distinction that matters most is whether the tool works from your actual, connected positions or reasons from something else: a separate account it manages for you, or the tickers and numbers you feed it. It decides how personal the answers can be, and where your money physically lives.

• Walnut: reads your real connected holdings. Walnut connects your real brokerage (Yes (the brokerage you already use, read-only by default)) and works from your actual positions, so its answers reflect what you genuinely own rather than a generic model.
• Alpaca: connects specific supported accounts. Alpaca connects a defined set of accounts (Alpaca accounts), so whether it can see your holdings depends on whether your money is at one of them.

This is where they diverge most. One works from your real, connected account while the other does not, so if you want an assistant grounded in the exact positions you already hold, that gap is the thing to weigh first. This holdings-aware angle is the one Walnut is built around: it connects the brokerage you already use and reasons from your live positions, read-only by default, with any trades left for you to approve.

Each one wins a different job, so the answer is committed per job rather than overall.

• Choose Walnut if you want a keep, trim or exit read on the stocks you already own, and placing the trades you approve. Its AI reads your real holdings, rates each one against the S&P 500, builds baskets and drafts trades, and it costs free; Walnut Unlimited $9.99/mo on the web; no fee on assets. It is built for investors who already own stocks at a broker and want an AI investing app to tell them which to keep, trim or exit, without moving their money. Keep in mind: it has no cost basis or tax lots, so returns are one-year price returns rather than profit and loss, and it is read-only at Schwab, Fidelity and Vanguard.
• Choose Alpaca if you want developers building AI agents. Its AI official MCP server (data + execution), and it costs free (open source). It is built for developers building their own automated or AI-driven trading systems. Keep in mind: developer-first; requires code and self-hosting.

Because they sit in different categories, this is not strictly either-or: some investors use one for a keep, trim or exit read on the stocks you already own, and placing the trades you approve and the other for developers building AI agents, and watch for overlapping costs.

Cost is easy to misread when two tools charge in different shapes, so compare the model, not just the number. Walnut: free; Walnut Unlimited $9.99/mo on the web; no fee on assets. Alpaca: free (open source). A percentage-of-assets fee scales with your balance, a flat subscription does not, and a “free” tier usually earns elsewhere (on cash, order flow, or premium upgrades), so the cheapest headline is not always the cheapest outcome for your situation.

Prices checked October 2026. Pricing and tiers change, so confirm the current numbers on each provider’s own site before you decide.