cvc5 vs Z3

cvc5 is a free, open-source solver for satisfiability modulo theories problems. It determines whether first-order formulas are satisfiable or valid across combinations of background theories, including arrays, bit-vectors, floating-point numbers, arithmetic, and strings. It also has a Syntax-Guided Synthesis engine for synthesizing functions against those theories.

The solver can run as a standalone command-line tool or be used as a library, with C++, C, Java, and Python APIs. Its binary accepts SMT-LIB v2 and SyGuS-IF input, and it can produce proofs and unsatisfiable cores to explain unsatisfiability results. The project links to an online interface for running SMT-LIB scripts and provides source releases and binaries for Linux, macOS, and Windows. cvc5 is led jointly by Stanford University and the University of Iowa. Research and commercial use are described as essentially unrestricted. The default license is modified BSD; using optional GPLv3 dependencies makes the combined work subject to GPLv3.

Z3 is a theorem prover and satisfiability modulo theories (SMT) solver used in software verification and analysis. SMT-LIB2 is its default input format, and it supports the SMT-LIB format. The project documents bindings or APIs for C, C++, .NET, Java, Go, OCaml, Python, Julia, and JavaScript/TypeScript. Z3 can be built with CMake, vcpkg, or Bazel, and the repository links to pre-built binaries for stable and nightly releases. A project wiki also links to a page for trying Z3 in a browser. It supports proof artifacts and counterexamples, and listed input languages include SMT-LIB2, Rust, Smalltalk, and several of the languages with documented APIs. Z3 is MIT licensed, with downloads and source code available at no charge. Python is required to build it; additional toolchains are needed for Java, .NET, OCaml, and Julia APIs. The README says MSVC builds enable Control Flow Guard and Address Space Layout Randomization by default.