diff --git a/CHANGELOG.md b/CHANGELOG.md
index dc493ef..cc8b818 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,17 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.13.2] - 2026-09-14
+
+### Fixed
+- **The codex `small` and `medium` tiers run under a ChatGPT sign-in.** Both
+  resolved to `gpt-5.4-mini`, which codex rejects on every call when signed in
+  with a ChatGPT account: `The 'gpt-5.4-mini' model is not supported when using
+  Codex with a ChatGPT account.` They now resolve to `gpt-5.6-luna:low` and
+  `gpt-5.6-luna:medium`. `gpt-5.6-luna` is in the model list codex fetches for a
+  ChatGPT account and in the catalog bundled with the CLI for API-key auth, and
+  both mark it `supported_in_api`. `large` stays `gpt-5.5:medium`.
+
 ## [0.13.1] - 2026-09-14
 
 ### Fixed
diff --git a/conformance/vectors/capabilities/capabilities.json b/conformance/vectors/capabilities/capabilities.json
index a37c803..7cd9bb2 100644
--- a/conformance/vectors/capabilities/capabilities.json
+++ b/conformance/vectors/capabilities/capabilities.json
@@ -63,8 +63,8 @@
       },
       "codex": {
         "large": "gpt-5.5:medium",
-        "medium": "gpt-5.4-mini:medium",
-        "small": "gpt-5.4-mini:low"
+        "medium": "gpt-5.6-luna:medium",
+        "small": "gpt-5.6-luna:low"
       },
       "gemini": {
         "large": "gemini-3-pro-preview",
diff --git a/conformance/vectors/plan/codex-agent.json b/conformance/vectors/plan/codex-agent.json
index ad775fd..949da24 100644
--- a/conformance/vectors/plan/codex-agent.json
+++ b/conformance/vectors/plan/codex-agent.json
@@ -5,7 +5,7 @@
     "provider": "codex",
     "spec": {
       "prompt": "hi",
-      "model": "gpt-5.4-mini",
+      "model": "gpt-5.6-luna",
       "agent": true,
       "isolated": true,
       "timeout": 180,
@@ -34,7 +34,7 @@
       "--color",
       "never",
       "--model",
-      "gpt-5.4-mini",
+      "gpt-5.6-luna",
       "-c",
       "service_tier=fast",
       "--ignore-user-config",
diff --git a/conformance/vectors/plan/codex-effort-suffix.json b/conformance/vectors/plan/codex-effort-suffix.json
index 4795baf..d6b54bb 100644
--- a/conformance/vectors/plan/codex-effort-suffix.json
+++ b/conformance/vectors/plan/codex-effort-suffix.json
@@ -5,7 +5,7 @@
     "provider": "codex",
     "spec": {
       "prompt": "hi",
-      "model": "gpt-5.4-mini:medium",
+      "model": "gpt-5.6-luna:medium",
       "agent": false,
       "isolated": true,
       "timeout": 180,
@@ -34,7 +34,7 @@
       "--color",
       "never",
       "--model",
-      "gpt-5.4-mini",
+      "gpt-5.6-luna",
       "-c",
       "model_reasoning_effort=medium",
       "-c",
diff --git a/rust/conformance-gen/src/cases.rs b/rust/conformance-gen/src/cases.rs
index 1cb1891..81657f4 100644
--- a/rust/conformance-gen/src/cases.rs
+++ b/rust/conformance-gen/src/cases.rs
@@ -363,13 +363,13 @@ fn plan_cases() -> Vec<Case> {
             "codex",
             RunSpec {
                 agent: true,
-                ..spec("hi", "gpt-5.4-mini")
+                ..spec("hi", "gpt-5.6-luna")
             },
         ),
         plan_case(
             "codex-effort-suffix",
             "codex",
-            spec("hi", "gpt-5.4-mini:medium"),
+            spec("hi", "gpt-5.6-luna:medium"),
         ),
         codex_cfg_case(
             "codex-service-tier-none",
diff --git a/rust/spawnllm-core/src/route.rs b/rust/spawnllm-core/src/route.rs
index 66855ec..d6f33ed 100644
--- a/rust/spawnllm-core/src/route.rs
+++ b/rust/spawnllm-core/src/route.rs
@@ -65,8 +65,8 @@ pub fn capabilities() -> Capabilities {
                 "codex",
                 ModelTiers {
                     large: Some("gpt-5.5:medium"),
-                    medium: Some("gpt-5.4-mini:medium"),
-                    small: Some("gpt-5.4-mini:low"),
+                    medium: Some("gpt-5.6-luna:medium"),
+                    small: Some("gpt-5.6-luna:low"),
                 },
             ),
             (
diff --git a/spawnllm/backends/codex.py b/spawnllm/backends/codex.py
index 7c6d30e..2e605ac 100644
--- a/spawnllm/backends/codex.py
+++ b/spawnllm/backends/codex.py
@@ -29,8 +29,8 @@ class CodexCliBackend(CliBackend):
     """
 
     models: ClassVar[dict[TModel, str]] = {
-        "small": "gpt-5.4-mini:low",
-        "medium": "gpt-5.4-mini:medium",
+        "small": "gpt-5.6-luna:low",
+        "medium": "gpt-5.6-luna:medium",
         "large": "gpt-5.5:medium",
     }
     provider: ClassVar[ProviderName] = "codex"
diff --git a/tests/test_backends.py b/tests/test_backends.py
index c1b972e..f105bbd 100644
--- a/tests/test_backends.py
+++ b/tests/test_backends.py
@@ -361,8 +361,8 @@ def test_claude(self) -> None:
 
     def test_codex(self) -> None:
         assert CodexCliBackend().models == {
-            "small": "gpt-5.4-mini:low",
-            "medium": "gpt-5.4-mini:medium",
+            "small": "gpt-5.6-luna:low",
+            "medium": "gpt-5.6-luna:medium",
             "large": "gpt-5.5:medium",
         }