7 best financial data and investment APIs for AI agents in 2026

AI is part of investment research now, and AI agents are doing more of the reading. Anthropic released ten agent templates for financial services on May 5, 2026, covering pitchbooks, KYC screening and month-end close. In a Mercer survey of 131 asset managers published May 21, 2026, 55% had integrated AI into at least one investment process.

The same survey named the main obstacle, with 69% of firms citing data quality or access as a material barrier. A model that can't reach the filing guesses the figure. LLM grounding, giving the model live, cited sources instead of its training memory, fixes that, and it starts with the API underneath.

These are the best financial data APIs I'd wire into an investment research stack or an AI agent today, across four layers: filings and web data, fundamentals, news, and market prices.

Financial data APIs return financial information as structured data, such as JSON, that a program or AI agent can use directly. Most fall into four groups: market data, fundamental data, news and ratings, and filings and alternative data from the public web.

• Market data APIs: real-time and historical prices, volume, OHLC bars and index levels.
• Fundamental data APIs: income statements, balance sheets, cash flows, ratios and earnings call transcripts.
• News and ratings APIs: market-moving headlines, press releases, analyst ratings and economic releases.
• Filings and alternative data APIs: SEC filings, 13F holdings, macro series, funding rounds, and signals read from public web pages such as pricing pages, job posts and investor relations sites.

Most research stacks combine at least two of these.

An investment API is any API built around investing, and the term covers three different products. Investment data APIs return prices, fundamentals, filings and news. Account aggregation APIs read a user's holdings and transactions from their brokerage accounts. Trading APIs place and manage orders. This post covers investment data APIs, the kind research teams and AI agents call.

If you're building a portfolio app, you may need all three, but they come from different vendors and solve different problems.

Choose a financial data API by matching it to the data your workflow reads, how fresh that data must be, and how much history you need. Then check the free plan's limits and whether your AI agent can reach the API through an MCP server or SDK.

• Coverage: asset classes, exchanges and regions. Many APIs are strongest on US equities.
• Freshness: real-time, 15-minute delayed or end of day. Real-time US market data usually sits on paid plans.
• History: how many years of prices and statements each plan includes.
• Limits: requests per minute or per day, which decide whether a batch job over hundreds of tickers finishes.
• AI agent access: an official MCP server or a plugin for Claude and ChatGPT saves you from writing a wrapper.

Finance teams build the same few jobs again and again. These are the most common ones built with a web data layer such as the Firecrawl API, MCP server or CLI:

• Agentic research: an AI agent reads filings, transcripts and news to answer a question about a company, and decides how many sources it needs as it goes. For agents that click through pages, see the guide to browser agents.
• Adverse media and KYC screening: compliance products search the open web for news about an entity and read the full articles, not snippets.
• Counterparty and transaction monitoring: risk teams gather public information on counterparties at high volume and flag changes.
• Watchlist research: quant and fundamental teams collect sources for hundreds of names in parallel, so concurrency matters more than single-call speed.

Firecrawl is the context API for AI agents, reaching financial data providers through Firecrawl Alexandria and turning any public web page or PDF into clean Markdown or structured data on one API key.

Some financial data already exists in structured form: a 10-Q, a quarterly income statement, a CPI series, an analyst rating change. Some exists only as a web page: an investor relations deck, a pricing page that changed overnight, a regulator's notice. Firecrawl covers both on one API, so an AI agent doesn't need a separate vendor for each layer.

Athena Intelligence builds an AI analyst for enterprise teams in finance, law and consulting on Firecrawl. As its founding platform engineer put it: "Without Firecrawl, we would lose a crucial data source that powers our platform."

Firecrawl Alexandria gives AI agents structured data from official providers and public sources through the same Firecrawl API, MCP server and CLI. An agent finds the right tool, reads its inputs and price, and calls it, so it gets the filing or the statement itself instead of a news story about it.

The finance category covers these jobs, among others:

The same category also holds central bank and statistics sources such as the Federal Reserve Board, the World Bank and the IMF. Finance is one part of the catalog:

Finding a tool and reading its inputs is free, and each call costs the Firecrawl credits shown on the tool. Firecrawl Alexandria isn't part of keyless access, so it needs a signed-in account or an API key, and some providers ask a team admin to accept their terms once before the first call.

You don't need code for any of this. Add the Firecrawl plugin in ChatGPT or the Firecrawl connector in Claude, sign in, and paste a prompt. Swap in your own tickers:

Use Firecrawl Alexandria to pull NVDA's last four quarterly income
statements from Fiscal.ai, then read the latest earnings call transcript.
Compare what management said about data center demand with the reported
numbers, and quote the analyst questions that pushed back hardest.
Tell me the Firecrawl credit price of each tool before you run it.

Use Firecrawl Alexandria to build a pre-market brief for AAPL, MSFT and
AMZN: Benzinga's Why Is It Moving notes from the last 24 hours, and
analyst upgrades or downgrades this week with prior and new price
targets. Newest first, one section per ticker, with links.

Use Firecrawl Alexandria to get Datadog's latest 10-Q from SEC EDGAR and
pull the risk factors section. Then check what professional investors'
fund letters say about the company in Fiscal.ai, and list each firm's
stance and the risks it names.

Use Firecrawl Alexandria to check the latest CPI and GDP figures from
FRED, the market-implied odds for the next Fed meeting from CME FedWatch,
and the newest CFTC positioning in Treasury futures. Summarize what the
three say together in five bullets.

Each of these jobs has a step-by-step guide in Firecrawl articles, written for both assistants, such as real-time stock news in Claude, tracking analyst upgrades and downgrades and finding startups that raised money.

Developers can run the same calls from the Python and Node.js SDKs, the other official SDKs, the REST API, the Firecrawl CLI or the MCP server. This Python example finds matching tools, then pulls Nvidia's latest filings from SEC EDGAR and US inflation from FRED:

pip install firecrawl-py

import os
from firecrawl import Firecrawl
 
firecrawl = Firecrawl(api_key=os.environ["FIRECRAWL_API_KEY"])
 
# Find matching finance data tools next to web results
found = firecrawl.search(
    "Nvidia 10-K filing and quarterly income statement",
    sources=["web", "alexandria"],
    limit=3,
)
for tool in found.tools or []:
    print(tool.provider, tool.capability)
 
# Nvidia's latest 10-K and 10-Q filings from SEC EDGAR
filings = firecrawl.scrape(
    alexandria={
        "provider": "sec-gov",
        "capability": "filings/filings",
        "options": {"ticker": "NVDA", "forms": ["10-K", "10-Q"], "limit": 3},
    }
)
for f in filings.alexandria[0].data["filings"]:
    print(f["form"], f["filing_date"], f["primary_document_url"])
 
# US CPI, year over year, from FRED
cpi = firecrawl.scrape(
    alexandria={
        "provider": "fred-stlouisfed-org",
        "capability": "economic-data/series_observations",
        "options": {"series_id": "CPIAUCSL", "units": "pc1", "observation_start": "2026-01-01"},
    }
)
for row in cpi.alexandria[0].data["observations"][-3:]:
    print(row["date"], row["value"])

Output (trimmed):

10-Q 2026-08-26 https://www.sec.gov/Archives/edgar/data/1045810/000104581026000075/nvda-20260726.htm
10-Q 2026-05-20 https://www.sec.gov/Archives/edgar/data/1045810/000104581026000052/nvda-20260426.htm
10-K 2026-02-25 https://www.sec.gov/Archives/edgar/data/1045810/000104581026000021/nvda-20260125.htm
2026-07-01 3.30386
2026-08-01 3.35302
2026-09-01 3.71725

For TypeScript tools that fetch the newest 8-K for a ticker, a FRED series or Treasury records, see the latest 8-K filing API walkthrough.

For everything that lives only on the web, the Firecrawl endpoints cover the full path from finding a source to watching it change:

• Firecrawl /search: finds fresh sources on the live web and can return full page content in the same call. Pass sources=["news"] for news results, or combine ["web", "news"]. See mastering the Firecrawl search endpoint and a Python web search API walkthrough.
• Firecrawl /scrape: turns any URL into Markdown or structured JSON, including JavaScript-heavy investor relations pages. The best web scraping APIs comparison covers other managed options.
• Firecrawl /parse: reads PDFs and documents such as annual reports, earnings decks and fund letters, with OCR for scanned pages. For other options, see the PDF parsers for RAG comparison.
• Firecrawl /crawl: collects every page of a company site or document library.
• Firecrawl /map: lists a site's URLs fast, useful for finding a company's investor relations pages before you crawl.
• Firecrawl /interact: handles clicks, forms and dynamic pages in a real browser.
• Firecrawl /monitor: alerts your pipeline when a page you watch changes, such as a pricing page or a press release list.

Firecrawl also works in no-code tools such as Lovable, n8n and Zapier. It's listed among the best MCP servers for developers. For teams with security reviews, 67% of Fortune 100 companies use Firecrawl, and the enterprise page and security docs cover the controls.

Best for: AI agents and research teams that need filings, statements, transcripts, macro data, news and public web pages through one API key, MCP server or CLI. More finance workflows are on the Firecrawl finance page and in the investment and finance docs.

Financial Modeling Prep is a fundamentals-focused API with financial statements, earnings call transcripts, 13F filings and price data.

FMP suits equity research tools that need more than prices. It covers statements by ticker and period, institutional holdings and SEC filings, and it documents an MCP server for AI agents.

• income-statement, balance-sheet-statement, cash-flow-statement: financial statements in JSON or CSV
• earning-call-transcript: earnings call transcripts by company and quarter
• institutional-ownership: 13F institutional holdings
• SEC filings search by company, plus financials parsed from filings
• discounted-cash-flow: precomputed DCF valuations
• economic-indicators: macro series alongside company data

Free plan: the Basic plan is $0 a month with 250 calls a day. Paid plans raise the limit to calls per minute and extend history up to 30+ years (pricing).

Limits: the free plan's daily cap is low for batch jobs, and the Starter plan includes up to 5 years of history.

Docs: site.financialmodelingprep.com/developer/docs

Alpha Vantage is a stock API with time series, technical indicators, company fundamentals and economic data, and a free key for most of its datasets.

Alpha Vantage is a common first API for quant developers and students because the free key works across most endpoints. It also runs an official MCP server for AI agents.

• TIME_SERIES_DAILY_ADJUSTED: daily prices adjusted for splits and dividends
• RSI, MACD, BBANDS: technical indicators computed from the time series
• OVERVIEW and EARNINGS: company fundamentals and earnings history
• NEWS_SENTIMENT: news articles with sentiment scores
• EARNINGS_CALL_TRANSCRIPT: earnings call transcripts
• REAL_GDP, CPI: economic indicators

Free plan: 25 API requests a day across most datasets (support page). Premium plans start at $49.99 a month for 75 requests a minute (premium page).

Limits: real-time and 15-minute delayed US market data are premium-only, and the daily cap on the free key rules out large batch jobs.

Docs: alphavantage.co/documentation

Finnhub is a market data API with real-time US quotes, company news, fundamentals and alternative data.

Finnhub's free plan includes real-time US quotes and a WebSocket stream. Its paid plans add global fundamentals, longer history, transcripts and estimates.

• Real-time quotes and a WebSocket stream for US stocks
• Company news and press releases
• Standardized financial statements and financials as reported
• Earnings call transcripts with downloadable audio
• SEC filings and a filings search API
• Analyst estimates and earnings surprises

Free plan: $0 a month with 60 API calls a minute for personal use, US coverage, a year of company news and a WebSocket for 50 symbols (pricing).

Limits: standardized statements and long price history sit on paid plans. The All-in-One plan is listed at $3,500 a month, billed annually.

Docs: finnhub.io/docs/api

Tiingo is a financial data API with end-of-day prices, IEX market data, a news API and crypto data on flat-rate plans.

Tiingo publishes its prices plainly: one plan for individuals and one for internal commercial use, with redistribution priced separately. Its pricing page lists over 110,000 global securities and 30+ years of price history.

• End-of-day prices: adjusted daily prices for US stocks, ETFs and mutual funds
• IEX data: intraday prices from IEX
• News API: articles tagged by ticker
• Crypto API: prices aggregated across exchanges

Free plan: the Starter plan is $0 a month for personal use. Power is $30 a month for individuals, and internal commercial use is $50 a month (pricing).

Limits: standard plans include 3 months of queryable news history, with longer history through sales. Fundamentals through the API are an add-on.

Docs: tiingo.com/documentation

Massive is a market data platform with real-time and historical data for US stocks, options, indices, currencies and futures, through REST, WebSockets and flat files.

Polygon.io renamed itself Massive on October 30, 2025, and existing API keys kept working. It suits trading systems and backtests that need tick-level trades, quotes and bulk history. Massive also publishes an open-source MCP server, rebuilt in March 2026.

• Aggregates: OHLCV bars from one second to daily
• Trades and quotes: trade-by-trade data on higher plans
• WebSockets: streaming market data
• Options, indices, currencies and futures: each on its own plans
• Flat files: bulk history for loading into a data warehouse

Free plan: Stocks Basic is $0 a month with 5 API calls a minute, 2 years of history and end-of-day data. Stocks Starter is $29 a month with unlimited calls, 15-minute delayed data, WebSockets and flat files (pricing).

Limits: real-time data and longer history sit on higher plans, and each asset class is priced separately.

Docs: massive.com/docs

OpenBB is an open-source data platform and a commercial workspace that connect market data, fundamentals and your own data sources for investment research.

OpenBB doesn't sell data. Its open-source Open Data Platform gives you one interface over the providers you configure, such as FMP or Tiingo. The commercial OpenBB Workspace adds dashboards, an App Marketplace and Workspace MCP, announced May 26, 2026.

• Open Data Platform: one interface over many data providers
• OpenBB Workspace: research dashboards that combine structured and unstructured data
• Workspace MCP: governed access for AI agents
• Snowflake Native App: runs OpenBB inside a Snowflake environment

Free plan: the Open Data Platform is open source. Workspace is a paid product.

Limits: you still need an API key and plan for each underlying provider you connect.

Docs: docs.openbb.co

Most research teams don't need all seven. A common split is Firecrawl for filings, statements, transcripts, macro data and web sources through one key, FMP or Alpha Vantage for broad fundamentals, Finnhub, Tiingo or Massive for prices and streaming, and OpenBB if you want one interface over several providers. Start with the layer that fills your biggest gap.

If prices and fundamentals are covered but your agent can't read filings, transcripts or the web, add Firecrawl next. Firecrawl Alexandria brings SEC EDGAR, Fiscal.ai, Benzinga, FRED and the rest of its finance providers into the same API key you use for Firecrawl /search and Firecrawl /scrape.

For more on building with these top financial data APIs, the best deep research APIs guide covers combining web extraction and structured data in agentic workflows. The OpenAI Agents and Firecrawl walkthrough wires these tools together for earnings analysis and financial news. The best news API comparison goes deeper on news sources, and the company funding data API benchmark tests 17 providers on funding round accuracy and freshness.

Try Firecrawl Alexandria in ChatGPT, or create a free account to get an API key with 1,000 Firecrawl credits a month.