How to get an options chain

The short way

yfinance will hand you a chain in four lines and no account:

import yfinance as yf

ticker = yf.Ticker("AAPL")
expiries = ticker.options                      # every expiry date, as strings
chain = ticker.option_chain(expiries[0])       # .calls and .puts, as DataFrames

ticker.options is the list of expiries; option_chain() takes one of them and returns calls and puts as separate DataFrames. Strikes, bid, ask, last, volume and open interest are there. What is also there is a 15-minute delay and no contract behind any of it — this reads Yahoo's undocumented endpoints, and both of those facts matter more for options than they do for stocks.

What the options are

A chain, free and delayed. yfinance as above. Good for screening, studying past behaviour and building an interface. Not something to put under a product, for the same reasons it is not for stock prices, plus one more: the delay is on the instrument whose price moves fastest.

A chain with greeks, cheaply. EODHD sells options chains with greeks as a marketplace add-on at $29.99 a month, on top of an API you may already be paying for. Watch the metering: on EODHD a call is a unit of currency rather than a request, and an options request costs ten of them against the 100,000 daily allowance.

A chain from a general-purpose API. Alpha Vantage carries end-of-day options on its $49.99 plan and does not carry realtime options until the $199.99 one — the clearest illustration on this page of what the underlying licence costs a vendor. Intrinio sells options prices with greeks and implied volatility alongside its XBRL fundamentals, from $150 a month for a personal seat with no display or redistribution rights.

A surface rather than a chain. ORATS is options-only and the product is the modelling: smoothed implied volatility and greeks, end-of-day back to 2007, from $199 a month for the delayed API and $299 for live chains once you have signed the exchange agreements. If what you need is a consistent surface across time rather than yesterday's quotes, this is a different purchase from everything above.

History as files. Cboe DataShop sells straight from the exchange, priced per dataset, date and symbol: one full-market day of the EOD summary is $50, or $70 with implied volatility and greeks attached; every OPRA print for a day with the NBBO at trade time is $132. Tick Data sells the same asset class by the symbol-year, and the comparison with its own equities rate card is the whole argument of this page — $32 for a symbol-year of US equity Level I against $1,000 for OPRA options.

No API at all. Market Chameleon is a browser product on 15-minute-delayed data with earnings-move statistics and multi-leg strategy screeners, one plan at $99 a month. It is on this list because a lot of people asking for a chain want an answer rather than an array, and writing the analysis yourself is the expensive part.

Where this breaks

OPRA. Every US listed options quote, from every exchange, reaches the world through one processor, and its fee schedule is one public document. There is no cheaper venue to buy from the way there is in equities, so the floor under every vendor above is the same filing. On the schedule dated 26 February 2026 the professional display device fee is $31.50 a month, the non-professional subscriber fee is $1.25 falling to $0.60 at volume, and redistribution is $1,500 a month — $650 if you are query-service only. That last line is the one that ends side projects: the moment a live options quote reaches somebody who is not you, it is redistribution.

What a free tier actually returns. Three things get quietly trimmed, and none of them is a bug:

• The delay. Fifteen minutes on an instrument whose quote is a function of an underlying that has moved since. The bid and ask you are reading are not the ones you would be filled against.
• The chain's width. Near-dated, near-the-money strikes rather than every expiry and every strike, because the full response for a liquid name is large.
• The greeks. They are absent or unreliable, because they are a calculation over a surface rather than a field in the feed, and building that surface is the part with a cost.

Coverage stops at the US border. Almost every card here that carries options is US-only, for the reason above: one processor with one published price, against a per-venue negotiation everywhere else.

History is not a bigger version of live. Options history is every strike of every expiry of every underlying, every day, which is why it is sold as files rather than streamed, and why FirstRate Data — which does sell options history — sells it end-of-day only, unadjusted for splits, with no intraday at any price.

If you outgrow this

When the delay starts costing you, the next question is not which vendor but which subscriber you are. Professional and non-professional are the exchange's categories, not the vendor's, and the gap between them on the OPRA schedule is wider than the gap between most vendors. Why real-time stock data is so expensive has the mechanics.

When you need the surface rather than the quotes, that is a modelling purchase. ORATS sells it; QuantLib lets you build it yourself and supplies no data at all.

When you want the flow rather than the chain, Unusual Whales sells the options tape and Greek exposure over REST and WebSocket, which is a different question about the same asset class.

Everything in the catalogue that touches options is on the options data page, and the APIs among them are in market data APIs.