fix: resolve codex model alias

hermes_cli/model_switch.py

@@ -112,7 +112,10 @@ MODEL_ALIASES: dict[str, ModelIdentity] = {
     # OpenAI
     "gpt5":      ModelIdentity("openai", "gpt-5"),
     "gpt":       ModelIdentity("openai", "gpt"),
-    "codex":     ModelIdentity("openai", "codex"),
+    # "codex" used to point at a now-vanished openai/codex slug. The live
+    # catalogs expose GPT-5 Codex family models instead, so resolve the alias
+    # to the actual codex-branded family.
+    "codex":     ModelIdentity("openai", "gpt-5.3-codex"),
     "o3":        ModelIdentity("openai", "o3"),
     "o4":        ModelIdentity("openai", "o4"),
 

tests/hermes_cli/test_model_switch_codex_alias.py

@@ -0,0 +1,73 @@
+"""Regression tests for the /model codex alias.
+
+The short alias used to point at a nonexistent openai/codex catalog entry,
+which made `/model codex` fail even though Codex models are available.
+"""
+
+from unittest.mock import patch
+
+from hermes_cli.model_switch import switch_model
+
+
+_MOCK_VALIDATION = {
+    "accepted": True,
+    "persist": True,
+    "recognized": True,
+    "message": None,
+}
+
+
+@patch("hermes_cli.model_switch.get_model_info", return_value=None)
+@patch("hermes_cli.model_switch.get_model_capabilities", return_value=None)
+@patch("hermes_cli.models.detect_provider_for_model", return_value=None)
+@patch("hermes_cli.models.validate_requested_model", return_value=_MOCK_VALIDATION)
+def test_codex_alias_resolves_on_openai_codex(
+    _validate,
+    _detect,
+    _capabilities,
+    _info,
+):
+    """/model codex should resolve to a real Codex model on openai-codex."""
+    with patch(
+        "hermes_cli.model_switch.list_provider_models",
+        return_value=["gpt-5.4-mini", "gpt-5.4", "gpt-5.3-codex", "gpt-5.2-codex"],
+    ):
+        result = switch_model(
+            raw_input="codex",
+            current_provider="openai-codex",
+            current_model="gpt-5.4",
+            current_base_url="https://chatgpt.com/backend-api/codex",
+            current_api_key="***",
+        )
+
+    assert result.success is True
+    assert result.target_provider == "openai-codex"
+    assert result.new_model == "gpt-5.3-codex"
+
+
+@patch("hermes_cli.model_switch.get_model_info", return_value=None)
+@patch("hermes_cli.model_switch.get_model_capabilities", return_value=None)
+@patch("hermes_cli.models.detect_provider_for_model", return_value=None)
+@patch("hermes_cli.models.validate_requested_model", return_value=_MOCK_VALIDATION)
+def test_codex_alias_resolves_on_openrouter(
+    _validate,
+    _detect,
+    _capabilities,
+    _info,
+):
+    """The alias should also work when switching through openrouter."""
+    with patch(
+        "hermes_cli.model_switch.list_provider_models",
+        return_value=["openai/gpt-5.4", "openai/gpt-5.4-mini", "openai/gpt-5.3-codex"],
+    ):
+        result = switch_model(
+            raw_input="codex",
+            current_provider="openrouter",
+            current_model="openai/gpt-5.4",
+            current_base_url="https://openrouter.ai/api/v1",
+            current_api_key="***",
+        )
+
+    assert result.success is True
+    assert result.target_provider == "openrouter"
+    assert result.new_model == "openai/gpt-5.3-codex"