Free — no account required

Citation verification
was always supposed
to be free.

KeyCite costs $500–800 per seat per month. The underlying opinions are public domain, written by judges on the public payroll. FreeCite makes citation verification free for every attorney — not as a trial, not as a freemium tier. Free.

Look up a citation → What it returns ↓

Federal courts 1951 to present. No login. No subscription.

The Alternative

What you've been paying for.

Citation verification has been a paywall for over a century. The data behind it is public domain.

Capability	KeyCite (Westlaw)	Shepard's (LexisNexis)	FreeCite
Current treatment status	✓	✓	✓
Full treatment history	✓	✓	✓
Treatment type per citing case	✓	✓	✓
Speaker attribution (majority vs. dissent)	✓	✓	✓
Published accuracy methodology	✗	✗	✓
Issue-level treatment	✓	✓	v2
Coverage	1800s–present	1800s–present	Federal 1951–present
Price	~$500–800/seat/month	~$400–600/seat/month	Free

FreeCite v1 covers federal courts 1951–present. Issue-level treatment and pre-1951 coverage are on the roadmap.

What You Get

Six things every citation lookup returns.

Not a link to the opinion. Structured facts about it.

Currency status

Is this case still good law? One answer: good law, overruled, questioned, or limited. Not a list of documents — a classification.

status: "OVERRULED"
good_law_through: "2024-06-28"

Treatment history

Every subsequent case that has cited this case, in chronological order. How the doctrine moved from the original decision to today.

cited_by: 184 cases
negative_treatment: 3

Treatment classification

For each citing case: what did the court do to this precedent? Affirmed, reversed, distinguished, questioned, followed, overruled, limited, criticized.

Loper Bright v. Raimondo (2024)
treatment: "OVERRULES"

Speaker attribution

Every treatment tagged as majority, concurrence, or dissent. A dissent criticizing a case is not negative treatment of its precedential value. We know the difference.

speaker: "MAJORITY"
// dissent criticism excluded

Confidence score

Every classification comes with a calibrated confidence score from the extraction pipeline. Low-confidence treatments are flagged, not silently included.

confidence: 0.97
// from gold-set evaluation

Full provenance

Every treatment record traces to its source: case name, citation, docket, court, date, paragraph. The document is always one click away.

citation: "603 U.S. 369 (2024)"
para: 42 · section: "II-B"

Example

Chevron deference: 40 years in five entries.

Enter 467 U.S. 837 and you get this.

1984

Established

Chevron U.S.A. v. Natural Resources Defense Council, 467 U.S. 837

Two-step deference framework established. Courts must defer to agency interpretations of ambiguous statutes if reasonable.

2001

Narrowed

United States v. Mead Corp., 533 U.S. 218

Chevron deference limited to formal agency adjudications and rulemakings. Informal guidance letters may receive only Skidmore deference.

2022

Questioned

West Virginia v. EPA, 597 U.S. 697

Major questions doctrine applied. Chevron deference unavailable for regulations of vast economic and political significance without clear congressional authorization.

2024

Overruled

Loper Bright Enterprises v. Raimondo, 603 U.S. 369

Chevron deference overruled. Courts must exercise independent judgment on questions of statutory interpretation. Marbury v. Madison, not Chevron, is the governing standard.

Three distinctions

What makes this different from Google Scholar.

Google Scholar shows you a list of citing cases. That is not citation verification.

Majority ≠ Dissent

Speaker attribution on every treatment.

A dissent that sharply criticizes a holding is not negative treatment. It has no precedential weight. FreeCite tags every treatment record with the judicial speaker — majority, concurrence, or dissent — so negative counts are not inflated by minority opinions.

Classified, Not Listed

Treatment type, not just citation count.

Knowing 184 cases cited Chevron tells you nothing. Knowing that 3 of those treatments are negative, and which specific ones and why, is what matters before filing. FreeCite classifies each treatment: affirmed, reversed, distinguished, questioned, followed, overruled, limited, criticized.

Published Accuracy

We publish our methodology. They don't.

KeyCite and Shepard's have never published treatment classification accuracy because they have never needed to — there was no alternative. FreeCite publishes: overall F1 target (>0.95), false negative rate on negative treatment (<2%), precision and recall by treatment type. Transparency is the competitive stance.

Why It's Free

The data was always supposed to be free.

Federal court opinions are public domain. They were written by judges on the public payroll, published in federal reporters at public expense, and made available in full text through PACER and CourtListener at no charge. The citation treatment relationships between cases — who cited whom and how — are derivable from those same public documents.

There is no legal or technical reason citation verification costs $500–800 per seat per month. The only reason it does is that Thomson Reuters and LexisNexis built the infrastructure to extract it before anyone else did and have kept it proprietary since.

FreeCite is free because the data belongs to everyone. That is the complete public explanation. Omniarch's business is the structured assertion graph that powers FreeCite. Citation verification is what the graph can do. The API is how builders access the graph programmatically.