Midpage API: What It Offers and Who It Fits

Short answer: Midpage is a legal research company that also sells its data to other software makers. Its API (a way for software to request data) searches and retrieves US court opinions, statutes, regulations, constitutions and agency guidance, and a hosted MCP server (a connection that lets an AI assistant call the data) and a database copy are two other ways in. Access starts with a request form, with one published price of $1,000 a month for prototypes and internal use and a custom licence for production. This guide uses only Midpage's own pages, read in early October 2026, and says so where a page is silent.

TL;DR

• Midpage sells two things. A research and drafting app for lawyers, and data access for companies that build legal products.
• The API is one service for cases and laws. One base address, one key, and endpoints for opinions, courts, statutes, regulations and guidance.
• Entry is a form, then a price. A $1,000 monthly tier covers prototypes and internal use. Anything customer-facing goes on a custom annual licence.
• Some things are not published. Midpage's pages give no numeric rate limits, and I found no public terms for production use.
• Ask about the numbers. Two Midpage pages quote different counts for statutes and regulations, so confirm the figure that goes in your contract.

What Midpage is

Midpage describes itself on its home page as "Legal Research Data for AI", offering cases, statutes, regulations and more, either as its own app or as a plugin inside Claude, ChatGPT and Perplexity. Its pricing page lists app plans for lawyers and law students. A footnote there says those plans do not include API access.

The data side has its own page, Legal Data for Legal Tech Companies. It offers three ways in: an API, a hosted MCP server, and a SQL database copy (a private replica of the whole collection that you query yourself). The page says the data is for "legal products", meaning software built by other companies.

A short history of Midpage

Midpage's company page says it was founded in 2022, is based in New York, and describes itself as "a legal research platform and data provider." The same page says founder and CEO Otto Zastrow started it "after working in AI and law at ETH Zurich," the Swiss technical university.

The legal press noticed it in 2025. On June 2, LawSites, a legal technology blog by journalist Bob Ambrogi, reported that Midpage had raised a $4 million seed round and launched Proposition Search, where you type a point of law and get the cases that most directly support it. Ambrogi wrote that "there is a strong argument to be made that Midpage is picking up the mantle from where Casetext left off." Casetext was an earlier legal research startup known for concept-based search. Here is Zastrow, in his launch post as quoted by LawSites: "We're committed to building the best experience for hardcore legal researchers who leave no stone unturned."

That helps place the API. The company page says "Read and research on Midpage, or access the same data through APIs and AI integrations," so the product lawyers use and the data sold to software makers share one collection.

What the API covers

The developer page gives these figures for the full collection:

• 16M+ US opinions, published and unpublished.
• 6M+ statutes, regulations, codes, constitutions and agency guidance.
• All federal appellate and district courts, all state appellate courts, and statutes and regulations from 50 states plus DC.

Midpage's separate data quality page quotes 4M+ for statutes, regulations and agency guidance, with the same 16M+ for opinions. The two pages may count differently. If the number matters to you, ask Midpage which definition it uses and put that in writing.

The documentation says one API covers both case law and laws, with a single base address and a single bearer key (a secret string you send with each request to prove who you are). The laws endpoints are search, get one document by citation or id, and a list of the available codes and registers. For laws the docs also say:

• Statutes and regulations come from GovInfo and official state repositories.
• Current law is the default. A document counts as current when it is not marked historical, its effective date has arrived, no newer version is in force, and it is not a repeal marker.
• Older versions need an API key with a separate historical entitlement, meaning an extra permission on the key. Without it, those requests return an error.
• Search returns up to 50 results a page and cannot page past the first 10,000.
• Where a source does not state an effective date, the field is empty.

The same docs describe a citator: a tool that shows whether later courts treated a case favourably or critically. That belongs to the case-law side.

Where the word "citator" comes from

A citator is older than software. LexisNexis's own history of Shepard's Citations says "Shepard's Citator was first published in 1873 by Frank Shepard, a salesman in Chicago." During his sales meetings, it says, he noticed that "every attorney was doing the same thing: keeping notes of case treatment in the margins." The first pamphlets covered Illinois. Later they became adhesive labels that lawyers stuck into their casebooks. LexisNexis says the service has been online since the 1980s and that it bought Shepard's in 1998.

The idea has stayed the same for 150 years: before you cite a case, find out whether later courts treated it favourably or critically. LexisNexis says a team of attorney-editors reads and analyzes decisions to decide that treatment. Midpage describes its AI citator as showing "how later courts have treated an opinion, with supporting reasoning and quotations." So the question for any citator, old or new, is how a label was decided and whether you can read the passage behind it. A label is a pointer. The lawyer still reads the later case.

How access works

Midpage's pages show three parts to getting access.

1. Request. The tier buttons on the data page, "Start building" and "Request access", both lead to a request form hosted on Notion. I could not read the form's contents.
2. Keys. The docs say to create and manage keys in Midpage's developer dashboard, and add: "If you need help setting up an organization eligible for API keys, contact support@midpage.ai." So keys are tied to an eligible organisation. The dashboard sits behind a security check that blocked my automated visit.
3. Licence. The prototyping tier is month to month. Production use moves to a Developer Licensing Agreement.

The two published tiers, from the data page:

What Midpage publishes, in one table

One small check for engineers: the docs list eight endpoints, including two for courts, while the downloadable OpenAPI document lists six paths. If you plan to generate a client from the spec, compare the two first. OpenAPI is the common format for files like this one. The OpenAPI Initiative, which maintains it under the Linux Foundation, says the specification was originally based on the Swagger Specification donated by SmartBear Software.

What Midpage says about AI use

Some background helps with those pages. Anthropic open-sourced MCP on November 25, 2024, calling it "a new standard for connecting AI assistants to the systems where data lives." On December 9, 2025 it said it was donating MCP to the Agentic AI Foundation. That is a fund under the Linux Foundation, co-founded by Anthropic, Block and OpenAI. The MCP site compares it to "a USB-C port for AI applications." One standard means a data vendor can build one server and have it work in several assistants. It also means the standard keeps moving. That is why a preview version, such as Midpage's version 4, deserves a question about change notice.

Midpage built the MCP server for agents, so its pages address AI directly. Reading them together gives this:

• The data page says the MCP server has "no data retention and no lock-in" and that Midpage requires "Zero Day Retention" from every model provider it uses.
• The security page covers the web app, plugins and integrations. It says Midpage does not use customer data to train or fine-tune AI models. It also says plugin and integration workflows "may still share submitted queries with model providers". It adds: "Model providers may retain submitted queries for up to 60 days."
• MCP version 4 is marked as a preview, so tool names and fields can change. Version 3 stays available as the stable contract.

Those statements sit on different pages, and the pages do not say which one governs the data licence, so ask. Midpage's public Website Terms of Service cover its web apps, plugins and other offerings that reference them, and I found no public terms written for the data licence.

Who it fits

Midpage is a natural fit if you are building a legal AI product that needs case law with a citator next to statutes and regulations, and you want a vendor that sells both from one key. The prototyping tier lets a small team test before signing a long contract. Teams that want a SQL copy of the full collection will have a conversation with sales either way.

It is a harder fit if you need a customer-facing launch quickly, since production use waits on a negotiated licence, or if you want to start building without filling in a request form.

Questions to ask before you choose

1. Which statute and regulation count applies to my contract: 6M+ or 4M+?
2. What are the numeric rate limits on the prototyping tier, and what changes on the annual licence?
3. Does the $1,000 tier include historical versions, or only current law?
4. What counts as "internal use" if my team's output reaches a client?
5. Which retention terms apply to the API and MCP: zero day, or up to 60 days at the model provider?
6. Is attribution to Midpage required on my product screens, and is it negotiable?
7. If a source for one state goes stale, how would I find out?
8. What happens to my integration when MCP version 4 leaves preview?
9. How is a citator label decided, and can I read the quoted passage behind it?

Question 7 applies to any vendor, and the ten questions post covers the rest of the checklist.

Where to read the comparisons

This page covers Midpage alone. For a line-by-line comparison of coverage, endpoints and access against other US legal data APIs, see the two-way Midpage comparison and the three-way comparison with OpenLaws. For a wider map of what a legal data API is, start with US law data: the complete guide or the 2026 roundup of legal data APIs.

If your product centres on statutes, regulations, court rules and agency guidance, Vaquill AI is built for that and retrieves by citation.

FAQ

What is the Midpage API?

It is the data service Midpage sells to companies that build legal products. It searches and retrieves US court opinions, statutes, regulations, constitutions and agency guidance through one base address and one key. Midpage also offers the same data through a hosted MCP server and a SQL database copy.

Does Midpage have a free API?

Midpage's pages do not describe a free API tier. The lowest published data price is $1,000 a month for prototyping and internal use. The app has a two-week free trial, but its pricing page says app plans do not include API access.

How much does the Midpage API cost?

The one published price is $1,000 a month, month to month, for prototypes and internal use, with attribution to Midpage required. Production use is a custom annual licence, and Midpage does not publish that price.

Does Midpage publish rate limits?

The data page calls the prototyping tier "rate limited". In the documentation pages I read, I found no numbers. Ask Midpage for the figures before you size a project.

Does Midpage cover case law as well as statutes?

Yes. Its pages list 16M+ opinions from federal appellate and district courts and state appellate courts. Statutes, regulations, constitutions and agency guidance sit in the same API.

Can I connect Midpage to an AI assistant?

Yes. Midpage runs a hosted MCP server, which works with MCP-capable clients. It accepts a shared API key or per-user sign-in. Version 4 is a preview and version 3 remains available as the stable contract.

Is the Midpage API the same as a Midpage subscription?

No. The app plans for lawyers and law students do not include API access, according to the pricing page. API access runs through the data page, with its own tiers and request form.

Where can I compare Midpage with other US legal data APIs?

Two existing comparisons cover coverage, endpoints and access side by side: the two-way Midpage guide and the three-way guide with OpenLaws.

For readers who want the primary-law data behind this guide, see the US primary law API.

Priyansh Khodiyar

Co-Founder & CTO

Priyansh leads engineering and AI at Vaquill AI: the pipelines that pull statutes, regulations and court rules from every US jurisdiction's official publisher, and the REST API, MCP server and open dataset that serve them.