diff --git a/code-rs/core/prompt_coder.md b/code-rs/core/prompt_coder.md
index 34c1f3dc568..9ebd63de4a5 100644
--- a/code-rs/core/prompt_coder.md
+++ b/code-rs/core/prompt_coder.md
@@ -60,7 +60,7 @@ agent {
     "task": "Implement JWT middleware (RS256) with key rotation and unit/integration tests. Preserve existing OAuth flows. Provide README usage snippet.",
     "context": "Service: services/api (Rust Axum). Secrets via env. CI: `cargo test --all`.",
     "files": ["services/api", "services/api/src", "services/api/Cargo.toml"],
-    "models": ["code-gpt-5.4","claude-sonnet-4.5","gemini-3-flash"],
+    "models": ["code-gpt-5.4","claude-sonnet-4.6","gemini-3-flash-preview"],
     "output": "Middleware + passing tests + README snippet",
     "write": true // Allow changes - will launch every agent in a separate worktree
   }
diff --git a/code-rs/core/src/agent_defaults.rs b/code-rs/core/src/agent_defaults.rs
index 5dd9f0411a6..51082a56843 100644
--- a/code-rs/core/src/agent_defaults.rs
+++ b/code-rs/core/src/agent_defaults.rs
@@ -39,14 +39,14 @@ pub const DEFAULT_AGENT_NAMES: &[&str] = &[
     "code-gpt-5.3-codex",
     "code-gpt-5.3-codex-spark",
     "claude-opus-4.6",
-    "gemini-3-pro",
+    "gemini-3.1-pro-preview",
     // Straightforward / cost-aware
     "code-gpt-5.1-codex-mini",
-    "claude-sonnet-4.5",
-    "gemini-3-flash",
+    "claude-sonnet-4.6",
+    "gemini-3-flash-preview",
     // Mixed/general and alternates
     "claude-haiku-4.5",
-    "qwen-3-coder",
+    "qwen3-coder-plus",
     "cloud-gpt-5.1-codex-max",
 ];
 
@@ -179,7 +179,7 @@ const AGENT_MODEL_SPECS: &[AgentModelSpec] = &[
         cli: "claude",
         read_only_args: CLAUDE_OPUS_READ_ONLY,
         write_args: CLAUDE_OPUS_WRITE,
-        model_args: &["--model", "opus"],
+        model_args: &["--model", "claude-opus-4-6"],
         description: "Higher-capacity Claude model for complex reasoning; use when you want the strongest Claude.",
         enabled_by_default: true,
         aliases: &["claude-opus", "claude-opus-4.1", "claude-opus-4.5"],
@@ -188,15 +188,15 @@ const AGENT_MODEL_SPECS: &[AgentModelSpec] = &[
         pro_only: false,
     },
     AgentModelSpec {
-        slug: "claude-sonnet-4.5",
+        slug: "claude-sonnet-4.6",
         family: "claude",
         cli: "claude",
         read_only_args: CLAUDE_SONNET_READ_ONLY,
         write_args: CLAUDE_SONNET_WRITE,
-        model_args: &["--model", "sonnet"],
+        model_args: &["--model", "claude-sonnet-4-6"],
         description: "Balanced Claude model for implementation and debugging; a solid default when you want Claude.",
         enabled_by_default: true,
-        aliases: &["claude", "claude-sonnet"],
+        aliases: &["claude", "claude-sonnet", "claude-sonnet-4.5"],
         gating_env: None,
         is_frontline: false,
         pro_only: false,
@@ -207,7 +207,7 @@ const AGENT_MODEL_SPECS: &[AgentModelSpec] = &[
         cli: "claude",
         read_only_args: CLAUDE_HAIKU_READ_ONLY,
         write_args: CLAUDE_HAIKU_WRITE,
-        model_args: &["--model", "haiku"],
+        model_args: &["--model", "claude-haiku-4-5"],
         description: "Fast Claude model for simple tasks, drafts, and quick iterations; pick when latency matters.",
         enabled_by_default: true,
         aliases: &["claude-haiku"],
@@ -216,15 +216,16 @@ const AGENT_MODEL_SPECS: &[AgentModelSpec] = &[
         pro_only: false,
     },
     AgentModelSpec {
-        slug: "gemini-3-pro",
+        slug: "gemini-3.1-pro-preview",
         family: "gemini",
         cli: "gemini",
         read_only_args: GEMINI_PRO_READ_ONLY,
         write_args: GEMINI_PRO_WRITE,
-        model_args: &["--model", "pro"],
-        description: "Higher-capacity Gemini model for harder tasks; use when gemini-3-flash misses details.",
+        model_args: &["--model", "gemini-3.1-pro-preview"],
+        description: "Higher-capacity Gemini preview for harder tasks; use when gemini-3-flash-preview misses details.",
         enabled_by_default: true,
         aliases: &[
+            "gemini-3-pro",
             "gemini-3-pro-preview",
             "gemini-3",
             "gemini3",
@@ -236,29 +237,34 @@ const AGENT_MODEL_SPECS: &[AgentModelSpec] = &[
         pro_only: false,
     },
     AgentModelSpec {
-        slug: "gemini-3-flash",
+        slug: "gemini-3-flash-preview",
         family: "gemini",
         cli: "gemini",
         read_only_args: GEMINI_FLASH_READ_ONLY,
         write_args: GEMINI_FLASH_WRITE,
-        model_args: &["--model", "flash"],
-        description: "Primary Gemini default for most tasks; fast and low-cost with near gemini-3-pro quality.",
+        model_args: &["--model", "gemini-3-flash-preview"],
+        description: "Primary Gemini preview default for most tasks; fast and low-cost with near gemini-3.1-pro-preview quality.",
         enabled_by_default: true,
-        aliases: &["gemini", "gemini-flash", "gemini-2.5-flash"],
+        aliases: &[
+            "gemini",
+            "gemini-flash",
+            "gemini-3-flash",
+            "gemini-2.5-flash",
+        ],
         gating_env: None,
         is_frontline: false,
         pro_only: false,
     },
     AgentModelSpec {
-        slug: "qwen-3-coder",
+        slug: "qwen3-coder-plus",
         family: "qwen",
         cli: "qwen",
         read_only_args: QWEN_3_CODER_READ_ONLY,
         write_args: QWEN_3_CODER_WRITE,
-        model_args: &["-m", "qwen-3-coder"],
+        model_args: &["-m", "qwen3-coder-plus"],
         description: "Fast and capable alternative; useful as a second opinion or for cross-checking.",
         enabled_by_default: true,
-        aliases: &["qwen", "qwen3"],
+        aliases: &["qwen", "qwen3", "qwen-3-coder"],
         gating_env: None,
         is_frontline: false,
         pro_only: false,
diff --git a/code-rs/core/src/agent_tool.rs b/code-rs/core/src/agent_tool.rs
index 7c3f02365da..7e057af569d 100644
--- a/code-rs/core/src/agent_tool.rs
+++ b/code-rs/core/src/agent_tool.rs
@@ -2417,7 +2417,7 @@ pub fn create_agent_tool(allowed_models: &[String]) -> OpenAiTool {
                 },
             }),
                 description: Some(
-                    "Optional array of model names (e.g., ['code-gpt-5.4','claude-sonnet-4.5','code-gpt-5.3-codex-spark','gemini-3-flash'])".to_string(),
+                    "Optional array of model names (e.g., ['code-gpt-5.4','claude-sonnet-4.6','code-gpt-5.3-codex-spark','gemini-3-flash-preview'])".to_string(),
                 ),
         },
     );
@@ -2975,7 +2975,7 @@ mod tests {
         }
 
         let cfg = AgentConfig {
-            name: "claude-sonnet-4.5".to_string(),
+            name: "claude-sonnet-4.6".to_string(),
             command: "claude".to_string(),
             args: Vec::new(),
             read_only: true,
@@ -2989,7 +2989,7 @@ mod tests {
 
         let output = execute_model_with_permissions(
             "agent-test",
-            "claude-sonnet-4.5",
+            "claude-sonnet-4.6",
             "ok",
             true,
             None,
diff --git a/code-rs/core/src/config.rs b/code-rs/core/src/config.rs
index cade56e5339..626b84d5e09 100644
--- a/code-rs/core/src/config.rs
+++ b/code-rs/core/src/config.rs
@@ -3083,9 +3083,9 @@ model_verbosity = "high"
 
         assert!(enabled_names.contains("code-gpt-5.3-codex"));
         assert!(enabled_names.contains("code-gpt-5.4"));
-        assert!(enabled_names.contains("claude-sonnet-4.5"));
-        assert!(enabled_names.contains("gemini-3-pro"));
-        assert!(enabled_names.contains("qwen-3-coder"));
+        assert!(enabled_names.contains("claude-sonnet-4.6"));
+        assert!(enabled_names.contains("gemini-3.1-pro-preview"));
+        assert!(enabled_names.contains("qwen3-coder-plus"));
         Ok(())
     }
 
@@ -3296,20 +3296,20 @@ mod agent_merge_tests {
     fn gemini_alias_and_canonical_dedupe_prefers_last_state() {
         let agents = vec![
             agent("gemini-2.5-pro", "gemini", true),
-            agent("gemini-3-pro", "gemini", false),
+            agent("gemini-3.1-pro-preview", "gemini", false),
         ];
         let merged = merge_with_default_agents(agents);
 
         let gemini = merged
             .iter()
-            .find(|a| a.name.eq_ignore_ascii_case("gemini-3-pro"))
+            .find(|a| a.name.eq_ignore_ascii_case("gemini-3.1-pro-preview"))
             .expect("gemini present");
 
         assert!(!gemini.enabled, "later canonical disable should win");
         assert_eq!(
             merged
                 .iter()
-                .filter(|a| a.name.eq_ignore_ascii_case("gemini-3-pro"))
+                .filter(|a| a.name.eq_ignore_ascii_case("gemini-3.1-pro-preview"))
                 .count(),
             1,
             "should dedupe gemini alias/canonical"
@@ -3319,21 +3319,21 @@ mod agent_merge_tests {
     #[test]
     fn gemini_alias_disable_overrides_prior_canonical_enable() {
         let agents = vec![
-            agent("gemini-3-pro", "gemini", true),
+            agent("gemini-3.1-pro-preview", "gemini", true),
             agent("gemini-2.5-pro", "gemini", false),
         ];
         let merged = merge_with_default_agents(agents);
 
         let gemini = merged
             .iter()
-            .find(|a| a.name.eq_ignore_ascii_case("gemini-3-pro"))
+            .find(|a| a.name.eq_ignore_ascii_case("gemini-3.1-pro-preview"))
             .expect("gemini present");
 
         assert!(!gemini.enabled, "later alias disable should win");
         assert_eq!(
             merged
                 .iter()
-                .filter(|a| a.name.eq_ignore_ascii_case("gemini-3-pro"))
+                .filter(|a| a.name.eq_ignore_ascii_case("gemini-3.1-pro-preview"))
                 .count(),
             1,
             "should dedupe gemini alias/canonical"
diff --git a/code-rs/core/src/config/validation.rs b/code-rs/core/src/config/validation.rs
index b37eb4bfae1..bf311fe5b73 100644
--- a/code-rs/core/src/config/validation.rs
+++ b/code-rs/core/src/config/validation.rs
@@ -199,14 +199,29 @@ fn upgrade_legacy_model_slug(slug: &str) -> Option<String> {
         return Some("claude-opus-4.6".to_string());
     }
 
-    // Upgrade Gemini 2.5 Pro to Gemini 3 Pro (or preview alias)
-    if slug.eq_ignore_ascii_case("gemini-2.5-pro") || slug.eq_ignore_ascii_case("gemini-3-pro-preview") {
-        return Some("gemini-3-pro".to_string());
+    // Upgrade Anthropic Sonnet 4.5 to 4.6.
+    if slug.eq_ignore_ascii_case("claude-sonnet-4.5") {
+        return Some("claude-sonnet-4.6".to_string());
     }
 
-    // Upgrade Gemini 2.5 Flash to Gemini 3 Flash
-    if slug.eq_ignore_ascii_case("gemini-2.5-flash") {
-        return Some("gemini-3-flash".to_string());
+    // Upgrade Gemini Pro slugs to the latest preview track.
+    if slug.eq_ignore_ascii_case("gemini-2.5-pro")
+        || slug.eq_ignore_ascii_case("gemini-3-pro")
+        || slug.eq_ignore_ascii_case("gemini-3-pro-preview")
+    {
+        return Some("gemini-3.1-pro-preview".to_string());
+    }
+
+    // Upgrade Gemini Flash slugs to the latest preview track.
+    if slug.eq_ignore_ascii_case("gemini-2.5-flash")
+        || slug.eq_ignore_ascii_case("gemini-3-flash")
+    {
+        return Some("gemini-3-flash-preview".to_string());
+    }
+
+    // Upgrade the older Qwen coder slug to the current plus line.
+    if slug.eq_ignore_ascii_case("qwen-3-coder") {
+        return Some("qwen3-coder-plus".to_string());
     }
 
     // Keep codex variants on their existing line; upgrades are surfaced via the
diff --git a/code-rs/core/src/slash_commands.rs b/code-rs/core/src/slash_commands.rs
index be4d4fc104b..e95f6b11b64 100644
--- a/code-rs/core/src/slash_commands.rs
+++ b/code-rs/core/src/slash_commands.rs
@@ -297,9 +297,9 @@ mod tests {
         let defaults = get_default_models();
         assert!(defaults.iter().any(|v| v == "code-gpt-5.4"));
         assert!(defaults.iter().any(|v| v == "code-gpt-5.3-codex"));
-        assert!(!defaults.iter().any(|v| v == "qwen-3-coder"));
-        assert!(!defaults.iter().any(|v| v == "gemini-3-flash"));
-        assert!(!defaults.iter().any(|v| v == "claude-sonnet-4.5"));
+        assert!(!defaults.iter().any(|v| v == "qwen3-coder-plus"));
+        assert!(!defaults.iter().any(|v| v == "gemini-3-flash-preview"));
+        assert!(!defaults.iter().any(|v| v == "claude-sonnet-4.6"));
 
         restore_var("PATH", orig_path);
         restore_var("HOME", orig_home);
@@ -334,7 +334,7 @@ mod tests {
         }
 
         let defaults = get_default_models();
-        assert!(defaults.iter().any(|v| v == "claude-sonnet-4.5"));
+        assert!(defaults.iter().any(|v| v == "claude-sonnet-4.6"));
 
         restore_var("PATH", orig_path);
         restore_var("HOME", orig_home);
diff --git a/code-rs/core/tests/gemini_model_args.rs b/code-rs/core/tests/gemini_model_args.rs
index 7d204a7c9eb..a25d129a9b7 100644
--- a/code-rs/core/tests/gemini_model_args.rs
+++ b/code-rs/core/tests/gemini_model_args.rs
@@ -2,19 +2,22 @@ use code_core::agent_defaults::agent_model_spec;
 
 #[test]
 fn gemini_specs_use_long_model_flag() {
-    let pro = agent_model_spec("gemini-3-pro").expect("spec present");
-    assert_eq!(pro.model_args, ["--model", "pro"]);
+    let pro = agent_model_spec("gemini-3.1-pro-preview").expect("spec present");
+    assert_eq!(pro.model_args, ["--model", "gemini-3.1-pro-preview"]);
 
     // The shorthand `gemini` is treated as the primary Gemini default.
     let primary = agent_model_spec("gemini").expect("alias present");
-    assert_eq!(primary.slug, "gemini-3-flash");
-    assert_eq!(primary.model_args, ["--model", "flash"]);
+    assert_eq!(primary.slug, "gemini-3-flash-preview");
+    assert_eq!(primary.model_args, ["--model", "gemini-3-flash-preview"]);
 
-    // Legacy shorthand and older slugs should resolve to the newest Gemini 3 Pro.
+    // Legacy shorthand and older slugs should resolve to the newest Gemini Pro preview.
     let legacy_pro = agent_model_spec("gemini-2.5-pro").expect("spec present via alias");
-    assert_eq!(legacy_pro.slug, "gemini-3-pro");
+    assert_eq!(legacy_pro.slug, "gemini-3.1-pro-preview");
+
+    let prior_preview = agent_model_spec("gemini-3-pro").expect("prior preview alias present");
+    assert_eq!(prior_preview.slug, "gemini-3.1-pro-preview");
 
     let legacy_flash = agent_model_spec("gemini-2.5-flash").expect("spec present via alias");
-    assert_eq!(legacy_flash.slug, "gemini-3-flash");
-    assert_eq!(legacy_flash.model_args, ["--model", "flash"]);
+    assert_eq!(legacy_flash.slug, "gemini-3-flash-preview");
+    assert_eq!(legacy_flash.model_args, ["--model", "gemini-3-flash-preview"]);
 }
diff --git a/code-rs/tui/src/bottom_pane/model_selection_view.rs b/code-rs/tui/src/bottom_pane/model_selection_view.rs
index fff7b8dc68f..eab099cb411 100644
--- a/code-rs/tui/src/bottom_pane/model_selection_view.rs
+++ b/code-rs/tui/src/bottom_pane/model_selection_view.rs
@@ -144,6 +144,11 @@ pub(crate) struct ModelSelectionView {
     current_effort: ReasoningEffort,
     current_service_tier: Option<ServiceTier>,
     current_context_mode: Option<ContextMode>,
+    baseline_service_tier: Option<ServiceTier>,
+    baseline_context_mode: Option<ContextMode>,
+    defer_session_mode_toggles_until_close: bool,
+    staged_service_tier: Option<Option<ServiceTier>>,
+    staged_context_mode: Option<Option<ContextMode>>,
     use_chat_model: bool,
     app_event_tx: AppEventSender,
     is_complete: bool,
@@ -191,6 +196,11 @@ impl ModelSelectionView {
             current_effort,
             current_service_tier,
             current_context_mode,
+            baseline_service_tier: current_service_tier,
+            baseline_context_mode: current_context_mode,
+            defer_session_mode_toggles_until_close: false,
+            staged_service_tier: None,
+            staged_context_mode: None,
             use_chat_model,
             app_event_tx,
             is_complete: false,
@@ -198,6 +208,35 @@ impl ModelSelectionView {
         }
     }
 
+    pub(crate) fn defer_session_mode_toggles_until_close(&mut self) {
+        self.defer_session_mode_toggles_until_close = true;
+    }
+
+    pub(crate) fn flush_deferred_session_updates(&mut self) {
+        if let Some(service_tier) = self.staged_service_tier.take() {
+            if service_tier != self.baseline_service_tier {
+                let _ = self
+                    .app_event_tx
+                    .send(AppEvent::UpdateServiceTierSelection { service_tier });
+                self.baseline_service_tier = service_tier;
+            }
+        }
+
+        if let Some(context_mode) = self.staged_context_mode.take() {
+            if context_mode != self.baseline_context_mode {
+                let _ = self
+                    .app_event_tx
+                    .send(AppEvent::UpdateSessionContextModeSelection { context_mode });
+                self.baseline_context_mode = context_mode;
+            }
+        }
+    }
+
+    fn should_defer_session_mode_toggles(&self) -> bool {
+        self.defer_session_mode_toggles_until_close
+            && matches!(self.target, ModelSelectionTarget::Session)
+    }
+
     pub(crate) fn update_presets(&mut self, presets: Vec<ModelPreset>) {
         let include_fast_mode = self.target.supports_fast_mode();
         let include_context_mode = self.target.supports_fast_mode();
@@ -439,9 +478,13 @@ impl ModelSelectionView {
                         Some(ServiceTier::Fast)
                     };
                     self.current_service_tier = next_service_tier;
-                    let _ = self.app_event_tx.send(AppEvent::UpdateServiceTierSelection {
-                        service_tier: next_service_tier,
-                    });
+                    if self.should_defer_session_mode_toggles() {
+                        self.staged_service_tier = Some(next_service_tier);
+                    } else {
+                        let _ = self.app_event_tx.send(AppEvent::UpdateServiceTierSelection {
+                            service_tier: next_service_tier,
+                        });
+                    }
                     return;
                 }
                 EntryKind::ContextMode => {
@@ -451,9 +494,13 @@ impl ModelSelectionView {
                         Some(ContextMode::Auto) => Some(ContextMode::Disabled),
                     };
                     self.current_context_mode = next_context_mode;
-                    let _ = self.app_event_tx.send(AppEvent::UpdateSessionContextModeSelection {
-                        context_mode: next_context_mode,
-                    });
+                    if self.should_defer_session_mode_toggles() {
+                        self.staged_context_mode = Some(next_context_mode);
+                    } else {
+                        let _ = self.app_event_tx.send(AppEvent::UpdateSessionContextModeSelection {
+                            context_mode: next_context_mode,
+                        });
+                    }
                     return;
                 }
                 EntryKind::FollowChat => {
@@ -1387,6 +1434,97 @@ mod tests {
         assert!(!view.is_complete());
     }
 
+    #[test]
+    fn deferred_context_mode_waits_until_close() {
+        let presets = vec![make_preset("gpt-5.4")];
+        let (tx, rx) = mpsc::channel::<AppEvent>();
+        let mut view = ModelSelectionView::new(
+            presets,
+            "gpt-5.4".to_string(),
+            ReasoningEffort::Low,
+            None,
+            Some(ContextMode::Auto),
+            false,
+            ModelSelectionTarget::Session,
+            AppEventSender::new(tx),
+        );
+        view.defer_session_mode_toggles_until_close();
+
+        let _ = view.handle_key_event_direct(KeyEvent::new(KeyCode::Down, KeyModifiers::NONE));
+        let _ = view.handle_key_event_direct(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
+
+        assert!(rx.try_recv().is_err(), "event should wait until close");
+
+        view.flush_deferred_session_updates();
+
+        let event = rx.try_recv().expect("context mode event after close");
+        assert!(matches!(
+            event,
+            AppEvent::UpdateSessionContextModeSelection {
+                context_mode: Some(ContextMode::Disabled)
+            }
+        ));
+        assert!(!view.is_complete());
+    }
+
+    #[test]
+    fn deferred_fast_mode_waits_until_close() {
+        let presets = vec![make_preset("gpt-5.4")];
+        let (tx, rx) = mpsc::channel::<AppEvent>();
+        let mut view = ModelSelectionView::new(
+            presets,
+            "gpt-5.4".to_string(),
+            ReasoningEffort::Low,
+            None,
+            None,
+            false,
+            ModelSelectionTarget::Session,
+            AppEventSender::new(tx),
+        );
+        view.defer_session_mode_toggles_until_close();
+
+        let _ = view.handle_key_event_direct(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
+
+        assert!(rx.try_recv().is_err(), "event should wait until close");
+
+        view.flush_deferred_session_updates();
+
+        let event = rx.try_recv().expect("service tier event after close");
+        assert!(matches!(
+            event,
+            AppEvent::UpdateServiceTierSelection {
+                service_tier: Some(ServiceTier::Fast)
+            }
+        ));
+        assert!(!view.is_complete());
+    }
+
+    #[test]
+    fn deferred_context_mode_skips_event_when_restored_before_close() {
+        let presets = vec![make_preset("gpt-5.4")];
+        let (tx, rx) = mpsc::channel::<AppEvent>();
+        let mut view = ModelSelectionView::new(
+            presets,
+            "gpt-5.4".to_string(),
+            ReasoningEffort::Low,
+            None,
+            Some(ContextMode::Auto),
+            false,
+            ModelSelectionTarget::Session,
+            AppEventSender::new(tx),
+        );
+        view.defer_session_mode_toggles_until_close();
+
+        let _ = view.handle_key_event_direct(KeyEvent::new(KeyCode::Down, KeyModifiers::NONE));
+        let _ = view.handle_key_event_direct(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
+        let _ = view.handle_key_event_direct(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
+        let _ = view.handle_key_event_direct(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
+
+        view.flush_deferred_session_updates();
+
+        assert!(rx.try_recv().is_err(), "restoring the original mode should emit nothing");
+    }
+
     #[test]
     fn model_selection_shows_unavailable_context_hint_for_unsupported_model() {
         let presets = vec![make_preset("gpt-5.3-codex")];
diff --git a/code-rs/tui/src/chatwidget.rs b/code-rs/tui/src/chatwidget.rs
index 61f31649761..4be9707a8a6 100644
--- a/code-rs/tui/src/chatwidget.rs
+++ b/code-rs/tui/src/chatwidget.rs
@@ -31618,7 +31618,7 @@ use code_core::protocol::OrderMeta;
         let (rows, _commands) = chat.collect_agents_overview_rows();
         let qwen = rows
             .iter()
-            .find(|row| row.name == "qwen-3-coder")
+            .find(|row| row.name == "qwen3-coder-plus")
             .expect("qwen row present");
         assert_eq!(qwen.enabled, qwen.installed);
 
@@ -33814,8 +33814,8 @@ use code_core::protocol::OrderMeta;
             write: false,
             write_requested: Some(false),
             models: Some(vec![
-                "claude-sonnet-4.5".to_string(),
-                "gemini-3-pro".to_string(),
+                "claude-sonnet-4.6".to_string(),
+                "gemini-3.1-pro-preview".to_string(),
             ]),
         }];
         chat.auto_state.pending_agent_timing = Some(AutoTurnAgentsTiming::Blocking);
@@ -33829,7 +33829,7 @@ use code_core::protocol::OrderMeta;
         assert!(message.contains("Run diagnostics"));
         assert!(message.contains("Please run agent.create"));
         assert!(message.contains("write: false"));
-        assert!(message.contains("Models: [claude-sonnet-4.5, gemini-3-pro]"));
+        assert!(message.contains("Models: [claude-sonnet-4.6, gemini-3.1-pro-preview]"));
         assert!(message.contains("Draft alternative fix"));
         assert!(message.contains("Focus on parser module"));
         assert!(message.contains("agent.wait"));
diff --git a/code-rs/tui/src/chatwidget/settings_overlay.rs b/code-rs/tui/src/chatwidget/settings_overlay.rs
index b4aea76d077..b662c390216 100644
--- a/code-rs/tui/src/chatwidget/settings_overlay.rs
+++ b/code-rs/tui/src/chatwidget/settings_overlay.rs
@@ -176,6 +176,8 @@ pub(crate) struct ModelSettingsContent {
 
 impl ModelSettingsContent {
     pub(crate) fn new(view: ModelSelectionView) -> Self {
+        let mut view = view;
+        view.defer_session_mode_toggles_until_close();
         Self { view }
     }
 }
@@ -192,6 +194,10 @@ impl SettingsContent for ModelSettingsContent {
     fn is_complete(&self) -> bool {
         self.view.is_complete()
     }
+
+    fn on_close(&mut self) {
+        self.view.flush_deferred_session_updates();
+    }
 }
 
 pub(crate) struct ThemeSettingsContent {
diff --git a/config.toml.example b/config.toml.example
index 7983e1a7945..5ff7281144c 100644
--- a/config.toml.example
+++ b/config.toml.example
@@ -43,7 +43,7 @@ read-only = false
 description = "Qwen Coder assistant (tracks latest default model)"
 # No default -m: let the CLI choose its current default model
 # To pin a model explicitly, export QWEN_MODEL or add here, e.g.:
-# args = ["-m", "qwen-3-coder", "-y"]
+# args = ["-m", "qwen3-coder-plus", "-y"]
 args = ["-y"]
 # Optional environment variables (either name works; mirrored automatically):
 # env = { QWEN_API_KEY = "your-key", DASHSCOPE_API_KEY = "your-key" }
@@ -55,7 +55,7 @@ enabled = true
 read-only = true
 description = "Qwen Coder assistant in read-only mode"
 # No -y in read-only; omit -m to track latest default. To pin:
-# args = ["-m", "qwen-3-coder"]
+# args = ["-m", "qwen3-coder-plus"]
 args = []
 
 [[agents]]
diff --git a/docs/agents.md b/docs/agents.md
index 7f813a1d799..b73c08617b9 100644
--- a/docs/agents.md
+++ b/docs/agents.md
@@ -19,9 +19,9 @@ instructions = "Preamble added to this agent’s prompt"
 Field recap: `name` (slug/alias), `command` (absolute paths ok), `args*` (RO/RW lists override base), `env`, `read_only`, `enabled`, optional `description` and `instructions`.
 
 ### Built-in defaults
-If no `[[agents]]` are configured, Code advertises built-ins (gated by env `CODE_ENABLE_CLOUD_AGENT_MODEL` for cloud variants): `code-gpt-5.4`, `code-gpt-5.3-codex`, `claude-opus-4.6`, `gemini-3-pro`, `code-gpt-5.1-codex-mini`, `claude-sonnet-4.5`, `gemini-3-flash`, `claude-haiku-4.5`, `qwen-3-coder`, `cloud-gpt-5.1-codex-max`. Built-ins strip any user `--model/-m` flags to avoid conflicts and inject their own.
+If no `[[agents]]` are configured, Code advertises built-ins (gated by env `CODE_ENABLE_CLOUD_AGENT_MODEL` for cloud variants): `code-gpt-5.4`, `code-gpt-5.3-codex`, `code-gpt-5.3-codex-spark`, `claude-opus-4.6`, `gemini-3.1-pro-preview`, `code-gpt-5.1-codex-mini`, `claude-sonnet-4.6`, `gemini-3-flash-preview`, `claude-haiku-4.5`, `qwen3-coder-plus`, `cloud-gpt-5.1-codex-max`. Built-ins strip any user `--model/-m` flags to avoid conflicts and inject their own.
 
-Tip: `gemini` resolves to `gemini-3-flash` (fast/cheap). Use `gemini-3-pro` when you want the higher-capacity Gemini option.
+Tip: `gemini` resolves to `gemini-3-flash-preview` (fast/cheap). Use `gemini-3.1-pro-preview` when you want the higher-capacity Gemini preview.
 
 ## Subagents (`[[subagents.commands]]`)
 ```toml
diff --git a/docs/config.md b/docs/config.md
index 6ce1e707dc1..cfdef2c0a98 100644
--- a/docs/config.md
+++ b/docs/config.md
@@ -198,7 +198,7 @@ approval_policy = "never"
 
 Use `[[agents]]` blocks to register additional CLI programs that Code can launch as peers. Each block maps a short `name` (referenced elsewhere in the config) to the command to execute, optional default flags, and environment variables.
 
-> **Note:** Built-in model slugs (for example `code-gpt-5.4`, `claude-sonnet-4.5`) automatically inject the correct `--model` or `-m` flag. To avoid conflicting arguments, Code strips any `--model`/`-m` flags you place in `args`, `args_read_only`, or `args_write` before launching the agent. If you need a new model variant, add a slug in `code-rs/core/src/agent_defaults.rs` (or set an environment variable consumed by the CLI) rather than pinning the flag here.
+> **Note:** Built-in model slugs (for example `code-gpt-5.4`, `claude-sonnet-4.6`) automatically inject the correct `--model` or `-m` flag. To avoid conflicting arguments, Code strips any `--model`/`-m` flags you place in `args`, `args_read_only`, or `args_write` before launching the agent. If you need a new model variant, add a slug in `code-rs/core/src/agent_defaults.rs` (or set an environment variable consumed by the CLI) rather than pinning the flag here.
 
 ```toml
 [[agents]]
@@ -449,7 +449,7 @@ tool_timeout_sec = 30
 
 Sub-agents are orchestrated helper workflows you can trigger with slash commands (for example `/plan`, `/solve`, `/code`). Each entry under `[[subagents.commands]]` defines the slash command name, whether spawned agents run in read-only mode, which `agents` to launch, and extra guidance for both the orchestrator (Code) and the individual agents.
 
-By default (when no `[[agents]]` are configured) Code advertises these model slugs for multi-agent runs: `code-gpt-5.4`, `code-gpt-5.3-codex`, `claude-opus-4.6`, `gemini-3-pro`, `code-gpt-5.1-codex-mini`, `claude-sonnet-4.5`, `gemini-3-flash`, `claude-haiku-4.5`, and `qwen-3-coder`. The cloud counterpart, `cloud-gpt-5.1-codex-max`, only appears when `CODE_ENABLE_CLOUD_AGENT_MODEL=1` is set. (`gemini` resolves to `gemini-3-flash`.) You can override the list by defining `[[agents]]` entries or by specifying `agents = [ … ]` on a given `[[subagents.commands]]` entry.
+By default (when no `[[agents]]` are configured) Code advertises these model slugs for multi-agent runs: `code-gpt-5.4`, `code-gpt-5.3-codex`, `code-gpt-5.3-codex-spark`, `claude-opus-4.6`, `gemini-3.1-pro-preview`, `code-gpt-5.1-codex-mini`, `claude-sonnet-4.6`, `gemini-3-flash-preview`, `claude-haiku-4.5`, and `qwen3-coder-plus`. The cloud counterpart, `cloud-gpt-5.1-codex-max`, only appears when `CODE_ENABLE_CLOUD_AGENT_MODEL=1` is set. (`gemini` resolves to `gemini-3-flash-preview`.) You can override the list by defining `[[agents]]` entries or by specifying `agents = [ … ]` on a given `[[subagents.commands]]` entry.
 
 ```toml
 [[subagents.commands]]
diff --git a/docs/slash-commands.md b/docs/slash-commands.md
index 3d7ab6fb7ab..18f167fc70b 100644
--- a/docs/slash-commands.md
+++ b/docs/slash-commands.md
@@ -117,9 +117,9 @@ Implementation Notes
   `code-rs/core/src/slash_commands.rs`.
   When no `[[agents]]` are configured, the orchestrator advertises the
   following model slugs to the LLM for multi-agent runs: `code-gpt-5.4`,
-  `code-gpt-5.3-codex`, `claude-opus-4.6`, `gemini-3-pro`,
-  `code-gpt-5.1-codex-mini`, `claude-sonnet-4.5`, `gemini-3-flash`,
-  `claude-haiku-4.5`, and `qwen-3-coder` (with
+  `code-gpt-5.3-codex`, `code-gpt-5.3-codex-spark`, `claude-opus-4.6`,
+  `gemini-3.1-pro-preview`, `code-gpt-5.1-codex-mini`, `claude-sonnet-4.6`,
+  `gemini-3-flash-preview`, `claude-haiku-4.5`, and `qwen3-coder-plus` (with
   `cloud-gpt-5.1-codex-max` gated by `CODE_ENABLE_CLOUD_AGENT_MODEL`). (`gemini`
-  resolves to `gemini-3-flash`.) You can replace or pin this set via
+  resolves to `gemini-3-flash-preview`.) You can replace or pin this set via
   `[[agents]]` or per-command `[[subagents.commands]].agents`.