We're glad to announce that Version 12.0 of Mathematica and the Wolfram Language is here—and it's a big one. This major update introduces several hundred new functions in a range of areas, including mathematics and geometry, visualization, audio and image processing, machine learning, and more, as well as provides a major overhaul to the Mac and Linux notebook interfaces and numerous systemwide performance enhancements. Some key new features include:

Mathematics and Geometry

• Convex optimization solvers such as LinearOptimization, QuadraticOptimization and SemidefiniteOptimization, with vector inequalities such as VectorGreater for modeling vector-valued variables
• Improved NDSolve support for nonlinear PDEs using FEM and automated idealized n-body interaction using NBodySimulation
• Use Around to describe uncertainty in measured values, with built-in support in visualization functions
• A host of new functions for generating, analyzing and manipulating polygons and polyhedra (RandomPolygon, RandomPolyhedron, PolygonDecomposition, etc.), with added support for holes/voids
• Access built-in axiomatic theories for use in symbolic proofs with AxiomaticTheory

Visualization

• GeometricScene and GeometricAssertion for visualizing Euclidean geometry and its theorems
• Visualize different components and properties of complex-valued data and functions with ComplexListPlot, ReImPlot and AbsArgPlot
• Represent geodetic vectors with GeoVector and visualize geodetic vector fields using GeoVectorPlot and GeoStreamPlot
• Updated default appearance for GraphPlot visualizations, as well as new plot themes and updated properties
• Create, analyze and modify chemical species with Molecule and visualize their 2D and 3D structures with MoleculePlot and MoleculePlot3D

Audio and Image Processing

• Color-processing functions such as FindMatchingColor, ImageRecolor and ColorDetect, all powered by the new ColorsNear wrapper to represent a color neighborhood
• Multi-language support for SpeechSynthesize, as well as improved discovery and searching of available voices through VoiceStyleData
• AudioAnnotate framework for annotating audio objects
• ShortTimeFourier for computing and storing properties of short-time Fourier transforms and InverseSpectrogram for approximating a signal from a magnitude spectrogram

Machine Learning and Neural Networks

• High-level recognition functions such as ImageCases, FacialFeatures, AudioIdentify and SpeechRecognize
• Additional pre-trained models, new layers, new encoders and decoders, and multi-GPU support for the neural network framework
• Learn a generalized probability distribution from arbitrary data using LearnDistribution
• Replace missing values in data by generated values using SynthesizeMissingValues and find anomalous examples in data using FindAnomalies

These are just a few of the highlights from this release, which also includes an experimental Wolfram Compiler, a comprehensive new Microcontroller Kit, a bidirectional Unity game engine interface and expanded knowledge and functionality for Entity and EntityStore.

You can find out more about these and other Version 12 additions by reading through Stephen Wolfram's launch-day blog post and looking at the Summary of New Features in 12.