Migrate GPT-5.4 uses to GPT-5.6 variants (#33173)

## What changed

- Hide `gpt-5.4` and `gpt-5.4-mini` from model selection and direct existing users to `gpt-5.6-terra` and `gpt-5.6-luna`, respectively.
- Use Terra for memory consolidation and Luna for memory extraction and rate-limit switch prompts.
- Keep Amazon Bedrock GPT-5 catalog entries visible when their bundled model metadata is hidden.

## Testing

- Add coverage that all Amazon Bedrock GPT-5 models remain visible and update TUI expectations for the new model choices.

GitOrigin-RevId: fd69c8d5fb6b510c77204df6b7ac3f07e67cded2

codex-rs/model-provider/src/amazon_bedrock/catalog.rs

@@ -5,6 +5,7 @@ use codex_model_provider_info::AMAZON_BEDROCK_GPT_5_6_SOL_MODEL_ID;
 use codex_model_provider_info::AMAZON_BEDROCK_GPT_5_6_TERRA_MODEL_ID;
 use codex_models_manager::bundled_models_response;
 use codex_protocol::openai_models::ModelInfo;
+use codex_protocol::openai_models::ModelVisibility;
 use codex_protocol::openai_models::ModelsResponse;
 use codex_protocol::openai_models::ReasoningEffort;
 use codex_protocol::openai_models::ReasoningEffortPreset;
@@ -75,6 +76,7 @@ fn gpt_5_bedrock_model(
     model.priority = priority;
     model.context_window = Some(GPT_5_BEDROCK_CONTEXT_WINDOW);
     model.max_context_window = Some(GPT_5_BEDROCK_CONTEXT_WINDOW);
+    model.visibility = ModelVisibility::List;
     model.availability_nux = None;
     model.upgrade = None;
     model
@@ -164,6 +166,15 @@ mod tests {
         }
     }
 
+    #[test]
+    fn gpt_5_bedrock_models_are_visible() {
+        let catalog = static_model_catalog();
+
+        for model in catalog.models {
+            assert_eq!(model.visibility, ModelVisibility::List);
+        }
+    }
+
     #[test]
     fn gpt_5_6_bedrock_models_use_variant_metadata_and_max_reasoning_effort() {
         let catalog = static_model_catalog();

codex-rs/model-provider/src/provider.rs

@@ -87,11 +87,11 @@ pub const DEFAULT_APPROVAL_REVIEW_PREFERRED_MODEL: &str = "codex-auto-review";
 
 /// Default model used for memory extraction when a provider does not require a
 /// backend-specific model ID.
-pub const DEFAULT_MEMORY_EXTRACTION_PREFERRED_MODEL: &str = "gpt-5.4-mini";
+pub const DEFAULT_MEMORY_EXTRACTION_PREFERRED_MODEL: &str = "gpt-5.6-luna";
 
 /// Default model used for memory consolidation when a provider does not require
 /// a backend-specific model ID.
-pub const DEFAULT_MEMORY_CONSOLIDATION_PREFERRED_MODEL: &str = "gpt-5.4";
+pub const DEFAULT_MEMORY_CONSOLIDATION_PREFERRED_MODEL: &str = "gpt-5.6-terra";
 
 /// Runtime provider abstraction used by model execution.
 ///

codex-rs/tui/src/chatwidget/rate_limits.rs

@@ -3,7 +3,7 @@
 use super::*;
 use codex_app_server_protocol::CodexErrorInfo as AppServerCodexErrorInfo;
 
-pub(super) const NUDGE_MODEL_SLUG: &str = "gpt-5.4-mini";
+pub(super) const NUDGE_MODEL_SLUG: &str = "gpt-5.6-luna";
 pub(super) const RATE_LIMIT_SWITCH_PROMPT_THRESHOLD: f64 = 90.0;
 pub(super) const RATE_LIMIT_SWITCH_PROMPT_VIEW_ID: &str = "rate-limit-switch-prompt";
 

codex-rs/tui/src/chatwidget/snapshots/codex_tui__chatwidget__tests__model_selection_popup.snap

@@ -10,10 +10,7 @@ expression: popup
   3. gpt-5.6-luna           Fast and affordable agentic coding model.
   4. gpt-5.5                Frontier model for complex coding, research, and
                             real-world work.
-  5. gpt-5.4                Strong model for everyday coding.
-  6. gpt-5.4-mini           Small, fast, and cost-efficient model for simpler
-                            coding tasks.
-› 7. gpt-5.2 (current)      Optimized for professional work and long-running
+› 5. gpt-5.2 (current)      Optimized for professional work and long-running
                             agents.
 
   Press enter to confirm or esc to go back

codex-rs/tui/src/chatwidget/snapshots/codex_tui__chatwidget__tests__rate_limit_switch_prompt_popup.snap

@@ -3,10 +3,10 @@ source: tui/src/chatwidget/tests/status_and_layout.rs
 expression: popup
 ---
   Approaching rate limits
-  Switch to gpt-5.4-mini for lower credit usage?
+  Switch to gpt-5.6-luna for lower credit usage?
 
-› 1. Switch to gpt-5.4-mini                 Small, fast, and cost-efficient
-                                            model for simpler coding tasks.
+› 1. Switch to gpt-5.6-luna                 Fast and affordable agentic coding
+                                            model.
   2. Keep current model
   3. Keep current model (never show again)  Hide future rate limit reminders
                                             about switching models.

codex-rs/tui/src/chatwidget/tests/popups_and_settings.rs

@@ -3675,7 +3675,7 @@ async fn advanced_only_reasoning_option_requires_explicit_selection() {
 #[tokio::test]
 async fn auto_model_advertising_advanced_effort_opens_reasoning_picker() {
     let (mut chat, mut rx, _op_rx) = make_chatwidget_manual(/*model_override*/ None).await;
-    let mut preset = get_available_model(&chat, "gpt-5.4");
+    let mut preset = get_available_model(&chat, "gpt-5.6-terra");
     preset.id = "codex-auto-test".to_string();
     preset.model = "codex-auto-test".to_string();
     preset.display_name = "codex-auto-test".to_string();

codex-rs/tui/src/chatwidget/tests/slash_commands.rs

@@ -401,7 +401,7 @@ async fn queued_slash_menu_cancel_drains_next_input() {
 async fn queued_settings_selection_applies_before_next_input() {
     let (mut chat, mut rx, mut op_rx) = make_chatwidget_manual(Some("gpt-5.2")).await;
     chat.thread_id = Some(ThreadId::new());
-    let mut preset = get_available_model(&chat, "gpt-5.4");
+    let mut preset = get_available_model(&chat, "gpt-5.6-terra");
     preset.supported_reasoning_efforts.truncate(1);
     let selected_effort = preset.supported_reasoning_efforts[0].effort.clone();
     chat.model_catalog = std::sync::Arc::new(ModelCatalog::new(vec![preset]));
@@ -439,7 +439,7 @@ async fn queued_settings_selection_applies_before_next_input() {
     match next_submit_op(&mut op_rx) {
         Op::UserTurn { model, effort, .. } => assert_eq!(
             (model, effort),
-            ("gpt-5.4".to_string(), Some(selected_effort))
+            ("gpt-5.6-terra".to_string(), Some(selected_effort))
         ),
         other => panic!("expected queued message with updated model, got {other:?}"),
     }