diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index b010363..b2a6898 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -9,10 +9,6 @@ name: release
 #    https://pypi.org/manage/project/geckopy/settings/publishing/ , pointing
 #    at this repo, workflow filename `release.yml`, and environment `pypi`.
 # 2. Create a `pypi` environment in this repo's Settings -> Environments.
-# Also note: geckopy's `raven-toolbox` dependency is currently a git direct
-# reference (see pyproject.toml), which PyPI rejects on upload -- the
-# pypi-publish job below will fail until that's switched to a released
-# version pin.
 
 on:
   release:
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index 747f59e..ef078fc 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -38,11 +38,5 @@ jobs:
       - name: Install
         run: pip install -e ".[dev,evotune]"
 
-      # ec_fseof needs the `equality` option of raven_toolbox.analysis.fseof,
-      # which is not in a release yet. Remove this step once geckopy requires
-      # a raven-toolbox release that has it.
-      - name: Install raven-toolbox from develop
-        run: pip install --upgrade --force-reinstall --no-deps "raven-toolbox @ git+https://github.com/CompMetEng/raven-toolbox.git@develop"
-
       - name: Run pytest
         run: pytest -q
diff --git a/CHANGELOG.md b/CHANGELOG.md
index c91849a..2b9ce06 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,6 +1,6 @@
 # Changelog
 
-## Unreleased
+## 4.0.0b2 — 2026-11-02
 
 * `pfba_enzymes` with `fraction_of_optimum` below 1 lets the objective fall by that fraction of the magnitude of its optimum, for a negative optimum and for a minimised objective as for a maximised positive one, as MATLAB's `getPfbaEnzymes` does; it raised `Infeasible` for these objectives.
 * `ec_fseof` ranks the targets by the score `|slope| * |correlation|` rounded to 9 significant digits, with scores that are equal after rounding in model order, as GECKO's `ecFSEOF` does; scores that differ by solver noise used to decide the order.
diff --git a/README.md b/README.md
index 3d7ab27..0c1cff1 100644
--- a/README.md
+++ b/README.md
@@ -16,7 +16,7 @@ Enzyme-constrained genome-scale metabolic modelling in Python: the Python port o
 
 An *ec*Model adds the limits of enzymes to a genome-scale model: each enzyme has a finite capacity (its kcat) and the cell has a finite protein budget. geckopy fetches kcats from BRENDA, predicts them with DLKcat, tunes them against growth and flux data with CMA-ES, applies custom curations and integrates proteomics.
 
-> **Status: beta (`4.0.0b1`).** The yeast-GEM tutorial runs end to end. The major and minor version follow GECKO, which is why it starts at 4.
+> **Status: beta (`4.0.0b2`).** The yeast-GEM tutorial runs end to end. The major and minor version follow GECKO, which is why it starts at 4.
 
 > **Not the older geckopy.** This is a new port of GECKO. It is unrelated to the earlier `geckopy` on PyPI (release 2.0.2 and earlier), described in Carrasco Muriel, Long & Sonnenschein, *Microbiology Spectrum* 11(6):e01705-23 (2023), [doi:10.1128/spectrum.01705-23](https://doi.org/10.1128/spectrum.01705-23).
 
diff --git a/pyproject.toml b/pyproject.toml
index 09393d0..efa246a 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "geckopy"
-version = "4.0.0b1"
+version = "4.0.0b2"
 description = "Enzyme-constrained genome-scale metabolic modeling in Python, inspired by the GECKO Toolbox"
 readme = "README.md"
 requires-python = ">=3.11"
@@ -44,10 +44,13 @@ dependencies = [
     # raven-toolbox provides the underlying model-manipulation primitives geckopy
     # extends, plus the ecModel YAML I/O (raven_toolbox.io.ec_data.EcData and the
     # ec-rxns / ec-enzymes / gecko_light sections in read/write_yaml_model).
-    # >=3.0.0b1: first release carrying convert_to_irreversible(..., rxns=...),
-    # which make_ec_model needs (#42). The "b1" in the specifier opts pip into
-    # matching pre-releases, so no --pre flag is needed downstream.
-    "raven-toolbox>=3.0.0b1",
+    # >=3.0.0b2: carries convert_to_irreversible(..., rxns=...), which
+    # make_ec_model needs (#42), raven_toolbox.utils.solver, which
+    # load_conventional_gem imports, and the `equality` option of
+    # raven_toolbox.analysis.fseof, which ec_fseof needs. The "b2" in the
+    # specifier opts pip into matching pre-releases, so no --pre flag is needed
+    # downstream.
+    "raven-toolbox>=3.0.0b2",
     "requests>=2.28",
     "ruamel.yaml>=0.17",
     "scipy>=1.10",