zhanghanduo · Oct 7, 2026 · Oct 7, 2026
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -7,6 +7,12 @@ the GitHub Release body, so a release with no entry here fails.
 
 Versioning follows [docs/versioning.md](docs/versioning.md).
 
+## [0.16.1] - 2026-10-07
+
+### Fixed
+
+- Runaway retry no longer expands the output cap past the model's limit when the client is driven through an `LLMFallbackChain`. The chain binds one `max_tokens` for whichever leg serves the request, so expansion is now clamped to the smallest declared output limit among its legs instead of reading a `capabilities` attribute the chain does not expose; a chain whose legs declare no limit keeps the previous context-bounded behaviour.
+
 ## [0.16.0] - 2026-10-07
 
 ### Changed

diff --git a/changes/74.fix.md b/changes/74.fix.md
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "apodex-agent-core"
-version = "0.16.0"
+version = "0.16.1"
 description = "Shared, product-neutral runtime primitives for Apodex agents"
 readme = "README.md"
 license = "Apache-2.0"

diff --git a/uv.lock b/uv.lock