Lean

Lean is a functional language and proof assistant for work that calls for machine-checked mathematics or verification. It is a strong fit for mathematicians, theorem provers, and developers prepared to formalize correctness; teams seeking a conventional programming environment should look elsewhere.

Overview

Lean brings programming and proof construction into one language, with a small trusted kernel at the center of its correctness claims. That makes it useful when a mathematical result or software property needs a proof the system can check, including work on hardware verification. It is a focused tool: its strengths matter most when formal assurance is part of the goal.

Created by Leonardo de Moura in 2013, Lean is maintained through the Lean FRO, a nonprofit formed under Convergent Research in July 2023. Its Apache 2.0 license permits free use and extension of the language and its mathematical libraries.

Key features

Mathlib contains more than a million lines of community-developed formal mathematics across algebra, analysis, topology, probability, and computer science. That gives formalization projects a substantial base to build on, especially in established areas, though it remains a specialized resource for work expressed in Lean.

Lean's Grind tool can automate involved pattern matching, case analysis, and systems of linear inequalities. This can shorten parts of proof construction, while metaprogramming lets users create domain-specific notation and proof automation. Those extension points reward people with the expertise to shape their own workflows.

Lean projects use Lake, the bundled package manager. The recommended setup uses VS Code and the Lean 4 extension for syntax highlighting and code completion. Neovim with lean.nvim is also well-supported; VSCodium and Cursor support is unofficial. For a first experiment, Lean4Web runs Lean code in a browser, and the learning resources include textbooks, tutorials, interactive games, core documentation, and an API reference.

A concrete verification use is AWS's work on Cedar, an authorization policy language: Lean is used to prove security properties and validate them against production code through differential testing. This illustrates a demanding application rather than a turnkey security product.

Pricing

Lean 4 costs nothing. The open-source language and proof assistant has no paid plans or free-trial clock, so individuals and organizations can use the same core software without a seat charge or usage quota stated here. Apache 2.0 also allows users to extend the code and mathematical libraries.

There are no lower tiers that withhold capabilities, but the free software does not include a paid service plan. Readers choosing Lean should weigh setup and formalization effort alongside its zero software price.

Platforms

Lean supports Linux, Apple Silicon macOS, and x86-64 Windows as Tier 1 platforms. x86-64 macOS and Emscripten WebAssembly are Tier 2, so users on those targets should account for their lower support tier. The browser option is useful for trying code, while the recommended desktop workflow centers on VS Code; platform support does not mean every editor has equal standing.

Who it's for

Lean is best for mathematicians formalizing results, theorem provers building machine-checked proofs, and programmers whose verification work justifies a formal language. The learning materials are organized to serve those three audiences, which helps newcomers choose a relevant route. General programmers can also use Lean, but should choose it when proof checking or formalization is a real requirement rather than a side benefit.

For collaboration and questions, the Lean Community Zulip is the central hub, and the Lean FRO hosts weekly technical office hours. These are useful avenues for users learning a specialized system, though they are community support rather than a commercial support plan.

Pros and cons

Pros: A minimal trusted kernel underpins proof checking, giving formal correctness work a clear foundation.

Pros: Mathlib's breadth and Grind's automation can support substantial mathematical formalization and reduce some repetitive proof steps.

Pros: The language, libraries, package manager, and editor integrations are available without a software fee, and projects can be extended through metaprogramming.

Cons: Lean's strongest case depends on willingness to formalize claims, which makes it a poor fit when conventional programming alone will do.

Cons: Users on x86-64 macOS or WebAssembly are on Tier 2, and VSCodium and Cursor support is unofficial.

Cons: The free software has no paid service plan; users seeking commercial support arrangements should account for that distinction.

Alternatives

Formal Verification Tools is the broader category for comparing approaches. Choose PVS if its free noncommercial plan suits the work; commercial users should note that its commercial pricing is not given here. Rocq is another free interactive theorem prover and dependently typed programming language, distributed under the GNU Lesser General Public License.

Pick cvc5 when a free open-source solver better matches the task; optional GPL-licensed dependencies affect the combined work's license. Why3 is a free option under the GNU Lesser General Public License 2.1 with a special linking exception. Z3 offers free MIT-licensed downloads and source code.

Boogie is free under the MIT license and requires.NET and a supported SMT solver. HOL4 is free software under the Modified BSD 3-Clause license. Alloy Analyzer is a free, open-source project distributed as a self-contained executable.

Verdict

Choose Lean if you need machine-checked proofs for mathematics or verification and want a programmable language backed by a large mathematical library. Its zero-cost, extensible toolchain is a persuasive foundation for formal work; look elsewhere if you need ordinary application development without the demands of proof formalization, or require a paid support plan.