Alpaca Review 2026 — The Developer-First Brokerage Everyone's Bolting AI Onto

There's a specific kind of company that quietly becomes critical infrastructure while most of the industry is watching something else. Alpaca is one of them.

If you've spent any time in the algorithmic trading, fintech, or AI-agent communities over the last three years, you've probably bumped into Alpaca as the default answer to "how do I actually execute a trade from code?" Commission-free stock, ETF, options, and crypto trading, exposed as a clean REST and streaming API, with a paper-trading mode that behaves exactly like live. Simple enough that a weekend hobbyist can wire up a trading bot. Serious enough that neo-brokers, robo-advisors, and embedded-finance startups build their entire product on top of it.

Then 2026 happened. Over the past four months Alpaca has shipped tokenized equities, 24/5 trading, zero-commission multi-leg options, a high-yield cash product, stock lending, an omnibus sub-accounting ledger, and — two days ago — version 2 of its official MCP Server that lets LLMs like Claude and ChatGPT place real trades in plain English.

That's a lot of product velocity for an industry famous for not having any. Worth a proper review.

What Alpaca actually is

Alpaca is an API-first brokerage built for developers, quant traders, and fintech companies. You get three core products that stack on each other:

• Trading API. Commission-free execution for US stocks, ETFs, options, and crypto, with a full paper-trading sandbox. This is the entry point for most individual users and the piece that powers most of the algorithmic-trading use cases.
• Market Data API. Real-time and historical data across equities, options, and crypto, with flexible timeframes, option Greeks, implied volatility, and corporate-actions feeds.
• Broker API. The white-label layer. If you're building a neo-broker, robo-advisor, employee equity platform, or any product that needs a brokerage underneath it, Broker API handles KYC, account opening, clearing, custody, and execution — so you can ship a brokerage product without becoming a regulated broker-dealer yourself.

That stack is why Alpaca shows up inside so many fintech products you'd never expect. The end-user-facing app looks like a shiny consumer brand; the plumbing underneath is Alpaca.

The target audience has always been clear: if you can write code, you're in. If you can't, you're better off with a retail brokerage like Robinhood, Schwab, or Trading 212. Alpaca has never pretended to be a consumer app.

The 2026 product drops, ranked by how much they matter

1. MCP Server v2 (April 15, 2026). The headline item. Alpaca's Model Context Protocol server lets AI assistants — Claude, ChatGPT, Cursor, VS Code, Gemini CLI — interact with the full trading stack in natural language. Research a stock, analyse a portfolio, construct a bull call spread, submit an order, all through a conversation. Version 2 is a complete FastMCP + OpenAPI rewrite, pushing tool count from 43 to 61 distinct actions. We covered this separately in the v2 deep-dive, but the short version is: this is the single most significant change to Alpaca's surface area in the last two years.

2. Instant Tokenization Network. Tokenized equities with 24/7 API access. This is the structural bet — representing real equities as on-chain assets that can be traded, settled, and collateralised outside traditional market hours. Still early, still regulated cautiously, still mostly interesting to institutional builders, but a genuine shift in what "a brokerage API" can mean. Worth watching even if you don't use it yet.

3. 24/5 Trading. Round-the-clock trading Monday through Friday, with overnight data. For AI-driven strategies that don't respect the 9:30-to-4 window, this matters more than it sounds. Overnight gaps are where retail gets hurt; 24/5 execution changes the risk profile of late-night event trades.

4. Multi-Leg Options. Spreads, straddles, and iron condors submitted as single combined orders with zero commissions. This is a genuine differentiator versus most API-first brokers, and it's what the MCP server's "bull call spread on AAPL" prompts are exercising underneath.

5. High Yield Cash. Competitive APY on uninvested balances with FDIC sweep coverage. Not revolutionary — most brokers have this now — but it closes a gap Alpaca had versus retail incumbents.

6. Stock Lending. Passive income on securities held in the account. Mostly a nice-to-have for retail; more interesting for the Broker API customers who want to offer this to their own users.

7. OmniSub (Sub-Accounting). An API-first ledger for omnibus clearing. Deeply in the weeds unless you're a Broker API customer, but a useful signal about where Alpaca is investing on the institutional side.

Pricing

Commission-free on the execution side. Revenue comes from market-data subscriptions, payment for order flow, margin lending, and the Broker API fee structure for embedded customers.

The paper-trading sandbox being free and unlimited is the big one. You can prototype anything — a day-trading bot, a robo-advisor, an AI-agent trading workflow — without paying a penny until you're ready to wire real money.

What's good

The developer experience is genuinely best-in-class. Clean REST, streaming WebSockets, official Python and JavaScript SDKs, thorough docs, a responsive status page. The onboarding gap between "I heard about Alpaca" and "I placed a paper trade from a script" is under an hour. That's rare in this category.

Paper trading that mirrors live. Same API, same endpoints, same behaviour — just different credentials. Most competitors' sandboxes feel like a parallel product with quirks of its own. Alpaca's doesn't.

The AI-agent posture is already ahead. No other developer-facing brokerage has shipped an official, first-class MCP server. Tradier, Tradestation, tastytrade — all decent APIs, none of them have native LLM tool exposure at this depth. If your roadmap involves AI agents, you're basically picking Alpaca or writing MCP glue yourself.

The Broker API is a serious piece of infrastructure. If you're a fintech, the list of things you don't have to build — KYC, account opening, custody, clearing, corporate actions, tax reporting — is long enough to change the viability of a whole category of product.

The 2026 velocity. Alpaca has shipped more material product in four months than most brokerage APIs ship in three years. That matters for a buying decision — you're picking a platform that's actively compounding, not coasting.

What's not great

US-only for retail. UK, EU, Australian, and most non-US retail users can only use paper trading directly. Live trading for non-US individuals requires either routing through a Broker API partner (some of which serve specific regions) or being a US resident. For UK readers specifically: you can build and test freely, you can't "open an Alpaca account and trade real money" without changing tax domicile.

The support model is lightweight. Alpaca is a developer product. Support is email/ticket-based with reasonable response times, but there's no phone line, no dedicated account manager unless you're on a serious Broker API contract. If you want hand-holding, pick a different broker.

Market-data costs escalate quickly. The free tier is generous for learning; the moment you need full SIP real-time data or extended history, you're in $99/month territory. Competitive with peers, but a surprise for people assuming "commission-free" means "free."

The retail UX is deliberately thin. No research tools, no screener UI, no portfolio analytics dashboard at the level of tastytrade or Interactive Brokers' Trader Workstation. Alpaca's answer is "build it yourself, or use the MCP server to talk to it." Fine if that's what you want. Frustrating if you expected a polished consumer product.

Some orders and order types still route conservatively. A handful of complex conditional orders and bracket variants behave differently than they would on institutional-grade platforms. Check the docs for edge cases before building around any exotic order logic.

Alpaca vs. Interactive Brokers vs. tastytrade vs. Tradier

Choose Alpaca if you're building AI-agent workflows, prototyping algorithmic strategies, or embedding brokerage into a product. Choose Interactive Brokers if you're a serious global trader who needs every market and asset class under one roof. Choose tastytrade if options are 80%+ of your strategy and you want best-in-class options tooling. Choose Tradier if Alpaca's approach appeals but you need a slightly different fee or data structure.

For the AI-tools audience this site exists for, the answer is mostly Alpaca, and the reason is almost entirely the MCP Server plus the Broker API combination.

Who should use it — and who should not

Use Alpaca if:

• You're a developer building a trading bot, algorithmic strategy, or AI-agent workflow and you want the lowest-friction path from "idea" to "paper order."
• You're a fintech company that needs a brokerage layer inside your product without becoming a regulated broker yourself.
• You're experimenting with LLM-driven financial workflows and want a platform where the MCP integration is first-class, not bolted on.
• You want to learn algorithmic trading with a real API, free, without risking capital.

Skip Alpaca if:

• You're a non-technical retail trader looking for a clean mobile app and research tools. Use Robinhood, Schwab, or Trading 212.
• You're a non-US resident looking to trade live with real money directly. Use Interactive Brokers or a regional alternative.
• Your strategy is options-heavy and you want the deepest single-vendor options experience. tastytrade is probably a better fit.
• You need markets outside the US — Asian equities, European exchanges, FX at scale. Interactive Brokers is the only serious answer.

How to get started

1. Sign up at alpaca.markets. Paper-trading accounts are instant. No KYC, no waiting, no minimum deposit. You get $100,000 of fake money to practice with.

2. Generate your API keys. Label them clearly — paper keys and live keys look similar and are catastrophic to confuse. Store them in a password manager, not a `.env` file committed to a repo.

3. Install the SDK. `pip install alpaca-py` for Python or the JavaScript SDK if that's your stack. Both are well-maintained.

4. Write your first order. The docs have a five-line "hello world" that places a paper buy order for one share of AAPL. Run it. Watch the dashboard update. That's the whole onboarding.

5. Add the MCP Server if you're AI-curious. Clone `alpacahq/alpaca-mcp-server`, follow the config, wire it to Claude Desktop or Cursor. Ten minutes later you have an LLM that can read your positions and simulate trades.

6. Only switch to live when you've proven the strategy in paper. Spend at least a few weeks in paper before wiring real money. This is good practice regardless of platform.

The bottom line

Alpaca isn't trying to be a consumer brokerage, and that's why developers love it. The API is clean, the paper trading mirrors live, the product velocity in 2026 has been genuine, and the MCP Server puts it meaningfully ahead of every other developer-facing broker on the LLM integration question.

It's not for everyone. US-only live trading, limited support, thin retail UX. If you need any of those things, look elsewhere.

But if you're the audience this site exists for — developers, quants, fintech builders, AI-tool operators — Alpaca is the boring, reliable choice that keeps getting more interesting. It's the kind of platform you build your career around, not because it's exciting, but because it doesn't let you down.

That's rare, and it's worth paying attention to when it happens.

Digital by Default reviews AI-native tools for UK operators and fintech builders. If you want help picking a brokerage or trading-infrastructure stack for your AI project, get in touch.