diff --git a/articles/underworld3-v3-2-0/figures/banner.png b/articles/underworld3-v3-2-0/figures/banner.png
new file mode 100644
index 0000000..9e2652c
Binary files /dev/null and b/articles/underworld3-v3-2-0/figures/banner.png differ
diff --git a/articles/underworld3-v3-2-0/metadata.yml b/articles/underworld3-v3-2-0/metadata.yml
new file mode 100644
index 0000000..6db3d27
--- /dev/null
+++ b/articles/underworld3-v3-2-0/metadata.yml
@@ -0,0 +1,31 @@
+# Validated in CI against schemas/article-metadata.schema.json.
+# `pixi run validate` checks this and the cross-file invariants a schema cannot
+# express -- that the article file is named <slug>.md, that canonical_path
+# matches the slug, and that no legacy DOI is ever paired with a new registrant.
+id: UWTN 2026-023
+slug: underworld3-v3-2-0
+title: "Underworld 3.2.0: dropping while it's hot!"
+article_type: release-note
+status: draft
+authors:
+  - name: Juan Carlos Graciosa
+    affiliation: Australian National University
+publication_date: null
+version: 1.0.0
+# The deposit writes archive_doi and repository_record_id when the note is
+# published; leave them out until then. `doi` and `doi_registrant` were here
+# once and are not fields the schema knows -- every note made from this
+# template failed `pixi run validate` on all three of them.
+license: CC-BY-4.0
+canonical_path: /underworld3-v3-2-0/
+legacy_paths: []
+# Facets, from vocabulary.yml. Both keys must be present even when empty: a
+# note with no subject is normal -- many are purely about method.
+subjects:
+methods:
+ghost_tags:
+  - Underworld Code
+banner: figures/banner.png
+banner_credit: null
+figures: 1
+source: native
diff --git a/articles/underworld3-v3-2-0/underworld3-v3-2-0.md b/articles/underworld3-v3-2-0/underworld3-v3-2-0.md
new file mode 100644
index 0000000..3ff385d
--- /dev/null
+++ b/articles/underworld3-v3-2-0/underworld3-v3-2-0.md
@@ -0,0 +1,84 @@
+---
+title: "Underworld 3.2.0: dropping while it's hot!"
+description: >-
+  The second quarterly release of Underworld3. Ten capabilities are newly
+  announced as supported, faults can be cut into a mesh as real discontinuities,
+  and the free surface runs in three dimensions on a spherical shell.
+date: 2026-10-14
+authors:
+  - name: Juan Carlos Graciosa
+    orcid: 0000-0003-0817-354X
+    affiliations:
+      - Australian National University
+license: CC-BY-4.0
+banner: figures/banner.png
+keywords:
+  - Underworld Code
+  - release
+  - development
+exports:
+  - format: typst
+    logo: ../../static/uwtn-logo.png
+    series: "Underworld Technical Notes"
+    origin_url: https://www.underworldcode.org/underworld3-v3-2-0/
+    template: ../../templates/pdf
+    output: underworld3-v3-2-0.pdf
+    article_id: UWTN 2026-023
+    article_version: 1.0.0
+    software_version: underworld3 v3.2.0
+---
+<div class="uwtn-banner"><img src="figures/banner.png" alt=""></div>
+
+Underworld 3.2.0 is out. Ten capabilities are newly announced as supported,
+seven as preview and one as experimental. 
+
+## What's new
+
+| Capability | What it does | Status | Documentation |
+|---|---|---|---|
+| **Analytic Benchmark Solutions** | Closed-form solutions for verifying solver convergence rates | Supported | [analytic](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/api/analytic.md) · [analytic-solutions](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/developer/subsystems/analytic-solutions.md) · [example: Ex_Stokes_Cartesian_SolC](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/examples/fluid_mechanics/advanced/Ex_Stokes_Cartesian_SolC.py) |
+| **Boundary Conditions on Curved Surfaces** | Free-slip and traction conditions using true geometric normals | Supported | [Free-slip boundary conditions on curved boundaries](/boundary-conditions-on-non-planar-boundaries/) · [example: Ex_Stokes_Annulus_Benchmark_Kramer](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/examples/fluid_mechanics/advanced/Ex_Stokes_Annulus_Benchmark_Kramer.py) |
+| **Checkpoint and Restart** | Snapshots, on-disk checkpoints and exact same-mesh reload | Supported | [snapshot-restore](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/snapshot-restore.md) · [create_xdmf](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/beginner/create_xdmf.md) |
+| **Constitutive Models** | Viscous, viscoplastic, viscoelastic-plastic, diffusion and Darcy laws | Supported (VEP failures fixed) | [Constitutive Models in Symbolic Form](/constitutive-models-in-symbolic-form/) · [example: vep-square](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/benchmarks/vep-square.md) |
+| **Geometric Multigrid / FMG Preconditioner** | Geometric full-multigrid preconditioning on refinement meshes | Supported (was Preview) | [Setting Up Full Multigrid](/setting-up-full-multigrid/) · [benchmark: FMG vs GAMG](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/figures/bench_fmg_vs_gamg_velocity_ksp.md) |
+| **Material History Transport** | Semi-Lagrangian, BDF and Adams-Moulton behind one D/Dt interface | Supported | [systems_ddt](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/api/systems_ddt.md) · [semi-lagrangian-time-integration](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/semi-lagrangian-time-integration.md) · [particle-population-and-materials](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/particle-population-and-materials.md) |
+| **Materials Registry and Distribution** | Separates what a material is from where it is | Supported | [materials](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/api/materials.md) · [example: particle-population-and-materials](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/particle-population-and-materials.md) |
+| **SUPG-Stabilised Transport** | SUPG stabilisation, the default for advection-diffusion and Navier-Stokes | Supported | [eulerian-navier-stokes](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/eulerian-navier-stokes.md) · [eulerian-advection-diffusion](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/eulerian-advection-diffusion.md) · [solvers](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/api/solvers.md) |
+| **Stress Tensor Visualisation** | Principal-stress glyphs and stress trajectories | Supported | [stress-visualisation](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/stress-visualisation.md) |
+| **Variable Data Access Contract** | One canonical write path for mesh and swarm variable data | Supported | [Mesh Variables and PETSc Vectors](/mesh-variables-and-petsc-vectors-keeping-arrays-in-sync/) · [example: 2-Variables](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/beginner/tutorials/2-Variables.ipynb) |
+| **Consistent Boundary Flux Recovery** | Pointwise surface flux recovered from the interior residual | Preview | [boundary_flux](https://github.com/underworldcode/underworld3/blob/v3.2.0/src/underworld3/utilities/boundary_flux.py) · [test_1019](https://github.com/underworldcode/underworld3/blob/v3.2.0/tests/test_1019_boundary_flux.py) |
+| **Fault Networks and Split-Node Faults** | Faults as real mesh discontinuities rather than smeared weak zones | Preview | [Faults: to mesh or not to mesh?](/putting-a-fault-in-a-mesh/) |
+| **Integration-Point Variables** | Material history stored at quadrature points rather than mesh nodes | Preview | [integration-point-variables](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/developer/subsystems/integration-point-variables.md) |
+| **Point Location and Function Evaluation** | Point location and symbolic evaluation in a distributed mesh | Preview | [Finding Particles in a Distributed, Unstructured Mesh](/finding-particles-in-a-distributed-unstructured-mesh/) · [example: 3-Symbolic_Forms](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/beginner/tutorials/3-Symbolic_Forms.ipynb) |
+| **Prognostic Free Surface** | Surface height relaxing toward stress-derived equilibrium topography | Preview | [Free surface in Underworld](/free-surface-in-underworld/) |
+| **Solver Control and Reporting** | Per-iteration callbacks, convergence reporting and stall detection | Preview | [solver-iteration-callbacks](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/solver-iteration-callbacks.md) |
+| **Transverse-Isotropic Rheology** | Anisotropic viscosity with a director | Preview (one accuracy drift unresolved) | [vep-transverse-isotropy-faults](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/vep-transverse-isotropy-faults.md) · [benchmark: bench_ti_vep_harmonic](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/benchmarks/bench_ti_vep_harmonic.py) |
+| **Geoid and Self-Gravity Response** | Geoid and self-gravity response for spherical shells and annuli | Experimental | [postprocessing](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/api/postprocessing.md) |
+
+## Four capabilities in more detail
+
+Geometric multigrid will matter a lot to those running big Stokes problems. If you have been letting Underworld fall back to algebraic multigrid without thinking about it, [Setting Up Full Multigrid](/setting-up-full-multigrid/) explains what the difference is worth. 
+
+Free-slip and traction conditions on curved boundaries are supported, which matters for modelling a lot of problems. Documentation is in [Boundary conditions on non-planar boundaries](/boundary-conditions-on-non-planar-boundaries/).
+
+SUPG stabilisation is now the default for both advection-diffusion and Navier-Stokes, so most models pick it up without being asked. Against the semi-Lagrangian alternative it is cheaper per step and much more accurate with small Courant numbers. Above a Courant number of about 2 the semi-Lagrangian scheme is better. [Eulerian Navier-Stokes](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/eulerian-navier-stokes.md) and [Eulerian advection-diffusion](https://github.com/underworldcode/underworld3/blob/v3.2.0/docs/advanced/eulerian-advection-diffusion.md) document the trade.
+
+The largest body of work this quarter went into faults. You can now specify a fault network once, as a trace and a hierarchy, and have it realised either as a split-node cut with contact laws or as a volumetric weak plane. Faults can now be modelled as real mesh discontinuities rather than smeared weak zones. [Faults: to mesh or not to mesh?](/putting-a-fault-in-a-mesh/) covers the choice and why it is not obvious.
+
+The prognostic free surface is now dimension-general and runs in three dimensions on a spherical shell. [A free surface in Underworld](/free-surface-in-underworld/) describes the method.
+
+## Things that will break
+
+There are five breaking API changes. They are listed with migration guidance in the release notes.
+
+## Getting it
+
+```bash
+git clone --branch v3.2.0 https://github.com/underworldcode/underworld3.git
+```
+
+Containers are at `ghcr.io/underworldcode/underworld3`, and if you would rather
+not install anything at all, you can run a notebook in your browser — see
+[Running Underworld in a Browser](/running-underworld-in-a-browser/).
+
+The full [release notes](https://github.com/underworldcode/underworld3/releases/tag/v3.2.0) list every bug fix, improvement and breaking change, and the names of everyone who contributed.