diff --git a/codex-rs/app-server-protocol/src/protocol/v2.rs b/codex-rs/app-server-protocol/src/protocol/v2.rs
index 963ac69000..ea019d3b61 100644
--- a/codex-rs/app-server-protocol/src/protocol/v2.rs
+++ b/codex-rs/app-server-protocol/src/protocol/v2.rs
@@ -26,7 +26,7 @@ use codex_protocol::config_types::ModeKind;
 use codex_protocol::config_types::Personality;
 use codex_protocol::config_types::ReasoningSummary;
 use codex_protocol::config_types::SandboxMode as CoreSandboxMode;
-use codex_protocol::config_types::ServiceTier;
+use codex_protocol::config_types::ServiceTier as CoreServiceTier;
 use codex_protocol::config_types::Verbosity;
 use codex_protocol::config_types::WebSearchMode;
 use codex_protocol::config_types::WebSearchToolConfig;
@@ -121,6 +121,47 @@ use serde_with::serde_as;
 use thiserror::Error;
 use ts_rs::TS;
 
+#[derive(Serialize, Deserialize, Debug, Clone, Copy, PartialEq, Eq, JsonSchema, TS)]
+#[serde(rename_all = "lowercase")]
+#[ts(export)]
+pub enum ServiceTier {
+    Fast,
+    Flex,
+}
+
+impl ServiceTier {
+    pub fn fast() -> Self {
+        Self::Fast
+    }
+
+    pub fn flex() -> Self {
+        Self::Flex
+    }
+
+    pub fn to_core(self) -> CoreServiceTier {
+        match self {
+            Self::Fast => CoreServiceTier::priority(),
+            Self::Flex => CoreServiceTier::flex(),
+        }
+    }
+
+    pub fn from_core(service_tier: &CoreServiceTier) -> Option<Self> {
+        if service_tier.is_priority() {
+            Some(Self::Fast)
+        } else if service_tier.is_flex() {
+            Some(Self::Flex)
+        } else {
+            None
+        }
+    }
+}
+
+impl From<ServiceTier> for CoreServiceTier {
+    fn from(value: ServiceTier) -> Self {
+        value.to_core()
+    }
+}
+
 // Macro to declare a camelCased API v2 enum mirroring a core enum which
 // tends to use either snake_case or kebab-case.
 macro_rules! v2_enum_from_core {
diff --git a/codex-rs/app-server/src/codex_message_processor.rs b/codex-rs/app-server/src/codex_message_processor.rs
index edc142c840..72dbf019d2 100644
--- a/codex-rs/app-server/src/codex_message_processor.rs
+++ b/codex-rs/app-server/src/codex_message_processor.rs
@@ -143,6 +143,7 @@ use codex_app_server_protocol::SendAddCreditsNudgeEmailParams;
 use codex_app_server_protocol::SendAddCreditsNudgeEmailResponse;
 use codex_app_server_protocol::ServerNotification;
 use codex_app_server_protocol::ServerRequestResolvedNotification;
+use codex_app_server_protocol::ServiceTier;
 use codex_app_server_protocol::SkillSummary;
 use codex_app_server_protocol::SkillsConfigWriteParams;
 use codex_app_server_protocol::SkillsConfigWriteResponse;
@@ -2538,7 +2539,7 @@ impl CodexMessageProcessor {
         let mut typesafe_overrides = self.build_thread_config_overrides(
             model,
             model_provider,
-            service_tier,
+            Self::resolve_service_tier_override(service_tier),
             cwd,
             approval_policy,
             approvals_reviewer,
@@ -2919,7 +2920,7 @@ impl CodexMessageProcessor {
                 thread: thread.clone(),
                 model: config_snapshot.model,
                 model_provider: config_snapshot.model_provider_id,
-                service_tier: config_snapshot.service_tier,
+                service_tier: config_snapshot.service_tier.as_ref().and_then(ServiceTier::from_core),
                 cwd: config_snapshot.cwd,
                 instruction_sources,
                 approval_policy: config_snapshot.approval_policy.into(),
@@ -3019,6 +3020,12 @@ impl CodexMessageProcessor {
         Ok(environment_selections)
     }
 
+    fn resolve_service_tier_override(
+        service_tier: Option<Option<ServiceTier>>,
+    ) -> Option<Option<codex_protocol::config_types::ServiceTier>> {
+        service_tier.map(|service_tier| service_tier.map(ServiceTier::to_core))
+    }
+
     async fn thread_archive(&self, request_id: ConnectionRequestId, params: ThreadArchiveParams) {
         let _thread_list_state_permit = match self.acquire_thread_list_state_permit().await {
             Ok(permit) => permit,
@@ -4381,7 +4388,7 @@ impl CodexMessageProcessor {
         let mut typesafe_overrides = self.build_thread_config_overrides(
             model,
             model_provider,
-            service_tier,
+            Self::resolve_service_tier_override(service_tier),
             cwd,
             approval_policy,
             approvals_reviewer,
@@ -4499,7 +4506,10 @@ impl CodexMessageProcessor {
                     thread,
                     model: session_configured.model,
                     model_provider: session_configured.model_provider_id,
-                    service_tier: session_configured.service_tier,
+                    service_tier: session_configured
+                        .service_tier
+                        .as_ref()
+                        .and_then(ServiceTier::from_core),
                     cwd: session_configured.cwd,
                     instruction_sources,
                     approval_policy: session_configured.approval_policy.into(),
@@ -5013,7 +5023,7 @@ impl CodexMessageProcessor {
             let mut typesafe_overrides = self.build_thread_config_overrides(
                 model,
                 model_provider,
-                service_tier,
+                Self::resolve_service_tier_override(service_tier),
                 cwd,
                 approval_policy,
                 approvals_reviewer,
@@ -5134,7 +5144,10 @@ impl CodexMessageProcessor {
                 thread: thread.clone(),
                 model: session_configured.model,
                 model_provider: session_configured.model_provider_id,
-                service_tier: session_configured.service_tier,
+                service_tier: session_configured
+                    .service_tier
+                    .as_ref()
+                    .and_then(ServiceTier::from_core),
                 cwd: session_configured.cwd,
                 instruction_sources,
                 approval_policy: session_configured.approval_policy.into(),
@@ -6769,7 +6782,7 @@ impl CodexMessageProcessor {
             let model = params.model;
             let effort = params.effort.map(Some);
             let summary = params.summary;
-            let service_tier = params.service_tier;
+            let service_tier = Self::resolve_service_tier_override(params.service_tier);
             let personality = params.personality;
 
             // If any overrides are provided, validate them synchronously so the
@@ -8492,7 +8505,7 @@ async fn handle_pending_thread_resume_request(
         thread,
         model,
         model_provider: model_provider_id,
-        service_tier,
+        service_tier: service_tier.as_ref().and_then(ServiceTier::from_core),
         cwd,
         instruction_sources,
         approval_policy: approval_policy.into(),
@@ -8662,11 +8675,15 @@ fn collect_resume_override_mismatches(
         ));
     }
     if let Some(requested_service_tier) = request.service_tier.as_ref()
-        && requested_service_tier != &config_snapshot.service_tier
+        && requested_service_tier.map(ServiceTier::to_core).as_ref()
+            != config_snapshot.service_tier.as_ref()
     {
+        let active_service_tier = config_snapshot
+            .service_tier
+            .as_ref()
+            .and_then(ServiceTier::from_core);
         mismatch_details.push(format!(
-            "service_tier requested={requested_service_tier:?} active={:?}",
-            config_snapshot.service_tier
+            "service_tier requested={requested_service_tier:?} active={active_service_tier:?}"
         ));
     }
     if let Some(requested_cwd) = request.cwd.as_deref() {
@@ -10048,6 +10065,7 @@ mod tests {
     use codex_model_provider_info::ModelProviderInfo;
     use codex_model_provider_info::WireApi;
     use codex_protocol::ThreadId;
+    use codex_protocol::config_types::ServiceTier as CoreServiceTier;
     use codex_protocol::openai_models::ReasoningEffort;
     use codex_protocol::permissions::FileSystemAccessMode;
     use codex_protocol::permissions::FileSystemPath;
@@ -10528,7 +10546,7 @@ mod tests {
             path: None,
             model: None,
             model_provider: None,
-            service_tier: Some(Some(codex_protocol::config_types::ServiceTier::Fast)),
+            service_tier: Some(Some(ServiceTier::fast())),
             cwd: None,
             approval_policy: None,
             approvals_reviewer: None,
@@ -10544,7 +10562,7 @@ mod tests {
         let config_snapshot = ThreadConfigSnapshot {
             model: "gpt-5".to_string(),
             model_provider_id: "openai".to_string(),
-            service_tier: Some(codex_protocol::config_types::ServiceTier::Flex),
+            service_tier: Some(CoreServiceTier::flex()),
             approval_policy: codex_protocol::protocol::AskForApproval::OnRequest,
             approvals_reviewer: codex_protocol::config_types::ApprovalsReviewer::User,
             permission_profile: codex_protocol::models::PermissionProfile::Disabled,
diff --git a/codex-rs/app-server/src/models.rs b/codex-rs/app-server/src/models.rs
index fd08098f74..74c4f2d066 100644
--- a/codex-rs/app-server/src/models.rs
+++ b/codex-rs/app-server/src/models.rs
@@ -6,6 +6,7 @@ use codex_app_server_protocol::ReasoningEffortOption;
 use codex_core::ThreadManager;
 use codex_models_manager::manager::RefreshStrategy;
 use codex_protocol::openai_models::ModelPreset;
+use codex_protocol::openai_models::ModelServiceTier as CoreModelServiceTier;
 use codex_protocol::openai_models::ReasoningEffortPreset;
 
 pub async fn supported_models(
@@ -22,6 +23,7 @@ pub async fn supported_models(
 }
 
 fn model_from_preset(preset: ModelPreset) -> Model {
+    let additional_speed_tiers = legacy_additional_speed_tiers(&preset.service_tiers);
     Model {
         id: preset.id.to_string(),
         model: preset.model.to_string(),
@@ -42,11 +44,18 @@ fn model_from_preset(preset: ModelPreset) -> Model {
         default_reasoning_effort: preset.default_reasoning_effort,
         input_modalities: preset.input_modalities,
         supports_personality: preset.supports_personality,
-        additional_speed_tiers: preset.additional_speed_tiers,
+        additional_speed_tiers,
         is_default: preset.is_default,
     }
 }
 
+fn legacy_additional_speed_tiers(service_tiers: &[CoreModelServiceTier]) -> Vec<String> {
+    service_tiers
+        .iter()
+        .filter_map(|tier| tier.id.is_priority().then_some("fast".to_string()))
+        .collect()
+}
+
 fn reasoning_efforts_from_preset(
     efforts: Vec<ReasoningEffortPreset>,
 ) -> Vec<ReasoningEffortOption> {
diff --git a/codex-rs/app-server/tests/common/models_cache.rs b/codex-rs/app-server/tests/common/models_cache.rs
index 3b4a58a7ab..af3e0da69f 100644
--- a/codex-rs/app-server/tests/common/models_cache.rs
+++ b/codex-rs/app-server/tests/common/models_cache.rs
@@ -28,7 +28,7 @@ fn preset_to_info(preset: &ModelPreset, priority: i32) -> ModelInfo {
         },
         supported_in_api: preset.supported_in_api,
         priority,
-        additional_speed_tiers: preset.additional_speed_tiers.clone(),
+        service_tiers: preset.service_tiers.clone(),
         upgrade: preset.upgrade.as_ref().map(Into::into),
         base_instructions: "base instructions".to_string(),
         model_messages: None,
diff --git a/codex-rs/app-server/tests/suite/v2/model_list.rs b/codex-rs/app-server/tests/suite/v2/model_list.rs
index 830ab0f783..ffb360e9c6 100644
--- a/codex-rs/app-server/tests/suite/v2/model_list.rs
+++ b/codex-rs/app-server/tests/suite/v2/model_list.rs
@@ -13,6 +13,7 @@ use codex_app_server_protocol::ModelUpgradeInfo;
 use codex_app_server_protocol::ReasoningEffortOption;
 use codex_app_server_protocol::RequestId;
 use codex_protocol::openai_models::ModelPreset;
+use codex_protocol::openai_models::ModelServiceTier as CoreModelServiceTier;
 use pretty_assertions::assert_eq;
 use tempfile::TempDir;
 use tokio::time::timeout;
@@ -50,11 +51,18 @@ fn model_from_preset(preset: &ModelPreset) -> Model {
         // cache report `supports_personality = false`.
         // todo(sayan): fix, maybe make roundtrip use ModelInfo only
         supports_personality: false,
-        additional_speed_tiers: preset.additional_speed_tiers.clone(),
+        additional_speed_tiers: legacy_additional_speed_tiers(&preset.service_tiers),
         is_default: preset.is_default,
     }
 }
 
+fn legacy_additional_speed_tiers(service_tiers: &[CoreModelServiceTier]) -> Vec<String> {
+    service_tiers
+        .iter()
+        .filter_map(|tier| tier.id.is_priority().then_some("fast".to_string()))
+        .collect()
+}
+
 fn expected_visible_models() -> Vec<Model> {
     // Filter by supported_in_api to support testing with both ChatGPT and non-ChatGPT auth modes.
     let mut presets = ModelPreset::filter_by_auth(
diff --git a/codex-rs/app-server/tests/suite/v2/thread_start.rs b/codex-rs/app-server/tests/suite/v2/thread_start.rs
index d8a50b88a4..cca4d085a6 100644
--- a/codex-rs/app-server/tests/suite/v2/thread_start.rs
+++ b/codex-rs/app-server/tests/suite/v2/thread_start.rs
@@ -14,6 +14,7 @@ use codex_app_server_protocol::McpServerStatusUpdatedNotification;
 use codex_app_server_protocol::RequestId;
 use codex_app_server_protocol::SandboxMode;
 use codex_app_server_protocol::ServerNotification;
+use codex_app_server_protocol::ServiceTier;
 use codex_app_server_protocol::ThreadStartParams;
 use codex_app_server_protocol::ThreadStartResponse;
 use codex_app_server_protocol::ThreadStartedNotification;
@@ -26,7 +27,6 @@ use codex_core::config::set_project_trust_level;
 use codex_exec_server::LOCAL_FS;
 use codex_git_utils::resolve_root_git_project_for_trust;
 use codex_login::REFRESH_TOKEN_URL_OVERRIDE_ENV_VAR;
-use codex_protocol::config_types::ServiceTier;
 use codex_protocol::config_types::TrustLevel;
 use codex_protocol::openai_models::ReasoningEffort;
 use pretty_assertions::assert_eq;
@@ -340,7 +340,7 @@ async fn thread_start_accepts_flex_service_tier() -> Result<()> {
 
     let req_id = mcp
         .send_thread_start_request(ThreadStartParams {
-            service_tier: Some(Some(ServiceTier::Flex)),
+            service_tier: Some(Some(ServiceTier::flex())),
             ..Default::default()
         })
         .await?;
@@ -352,7 +352,7 @@ async fn thread_start_accepts_flex_service_tier() -> Result<()> {
     .await??;
     let ThreadStartResponse { service_tier, .. } = to_response::<ThreadStartResponse>(resp)?;
 
-    assert_eq!(service_tier, Some(ServiceTier::Flex));
+    assert_eq!(service_tier, Some(ServiceTier::flex()));
     Ok(())
 }
 
diff --git a/codex-rs/codex-api/tests/models_integration.rs b/codex-rs/codex-api/tests/models_integration.rs
index 9f95c9441f..93e08af20d 100644
--- a/codex-rs/codex-api/tests/models_integration.rs
+++ b/codex-rs/codex-api/tests/models_integration.rs
@@ -74,7 +74,7 @@ async fn models_client_hits_models_endpoint() {
             visibility: ModelVisibility::List,
             supported_in_api: true,
             priority: 1,
-            additional_speed_tiers: Vec::new(),
+            service_tiers: Vec::new(),
             upgrade: None,
             base_instructions: "base instructions".to_string(),
             model_messages: None,
diff --git a/codex-rs/core/config.schema.json b/codex-rs/core/config.schema.json
index a30a3ed925..fa29de17fc 100644
--- a/codex-rs/core/config.schema.json
+++ b/codex-rs/core/config.schema.json
@@ -2151,10 +2151,7 @@
       "type": "object"
     },
     "ServiceTier": {
-      "enum": [
-        "fast",
-        "flex"
-      ],
+      "description": "String-backed service tier identifier. Known values include `priority` and `flex`, but other backend-provided ids are allowed.",
       "type": "string"
     },
     "ShellEnvironmentPolicyInherit": {
@@ -4497,4 +4494,4 @@
   },
   "title": "ConfigToml",
   "type": "object"
-}
\ No newline at end of file
+}
diff --git a/codex-rs/core/src/client.rs b/codex-rs/core/src/client.rs
index ba81b451a7..5ed1658d3d 100644
--- a/codex-rs/core/src/client.rs
+++ b/codex-rs/core/src/client.rs
@@ -889,11 +889,7 @@ impl ModelClientSession {
             store: provider.is_azure_responses_endpoint(),
             stream: true,
             include,
-            service_tier: match service_tier {
-                Some(ServiceTier::Fast) => Some("priority".to_string()),
-                Some(service_tier) => Some(service_tier.to_string()),
-                None => None,
-            },
+            service_tier: service_tier.map(|service_tier| service_tier.to_string()),
             prompt_cache_key,
             text,
             client_metadata: Some(HashMap::from([(
diff --git a/codex-rs/core/src/client_common_tests.rs b/codex-rs/core/src/client_common_tests.rs
index f67e4f1fd8..732ae63e4d 100644
--- a/codex-rs/core/src/client_common_tests.rs
+++ b/codex-rs/core/src/client_common_tests.rs
@@ -155,7 +155,7 @@ fn serializes_flex_service_tier_when_set() {
         stream: true,
         include: vec![],
         prompt_cache_key: None,
-        service_tier: Some(ServiceTier::Flex.to_string()),
+        service_tier: Some(ServiceTier::flex().to_string()),
         text: None,
         client_metadata: None,
     };
diff --git a/codex-rs/core/src/compact.rs b/codex-rs/core/src/compact.rs
index 58a2610fcb..9638086b78 100644
--- a/codex-rs/core/src/compact.rs
+++ b/codex-rs/core/src/compact.rs
@@ -518,7 +518,7 @@ async fn drain_to_completed(
             &turn_context.session_telemetry,
             turn_context.reasoning_effort,
             turn_context.reasoning_summary,
-            turn_context.config.service_tier,
+            turn_context.config.service_tier.clone(),
             turn_metadata_header,
             // Rollout tracing currently models remote compaction only; local compaction streams
             // are left untraced until the reducer has a first-class local compaction lifecycle.
diff --git a/codex-rs/core/src/config/mod.rs b/codex-rs/core/src/config/mod.rs
index 20b2a923f8..e6a5bf09a9 100644
--- a/codex-rs/core/src/config/mod.rs
+++ b/codex-rs/core/src/config/mod.rs
@@ -2659,12 +2659,10 @@ impl Config {
             None => config_profile.service_tier.or(cfg.service_tier),
         };
         let service_tier = match service_tier {
-            Some(ServiceTier::Fast) if features.enabled(Feature::FastMode) => {
-                Some(ServiceTier::Fast)
+            Some(service_tier) if service_tier.is_priority() && !features.enabled(Feature::FastMode) => {
+                None
             }
-            Some(ServiceTier::Fast) => None,
-            Some(ServiceTier::Flex) => Some(ServiceTier::Flex),
-            None => None,
+            other => other,
         };
 
         let compact_prompt = compact_prompt.or(cfg.compact_prompt).and_then(|value| {
diff --git a/codex-rs/core/src/session/mod.rs b/codex-rs/core/src/session/mod.rs
index c18976fde1..2393eab5b3 100644
--- a/codex-rs/core/src/session/mod.rs
+++ b/codex-rs/core/src/session/mod.rs
@@ -585,7 +585,7 @@ impl Codex {
             .auth_cached()
             .and_then(|auth| auth.account_plan_type());
         let service_tier = get_service_tier(
-            config.service_tier,
+            config.service_tier.clone(),
             config.notices.fast_default_opt_out.unwrap_or(false),
             account_plan_type,
             config.features.enabled(Feature::FastMode),
@@ -788,7 +788,7 @@ fn get_service_tier(
 
     account_plan_type
         .is_some_and(is_enterprise_default_service_tier_plan)
-        .then_some(ServiceTier::Fast)
+        .then(ServiceTier::priority)
 }
 
 fn is_enterprise_default_service_tier_plan(plan_type: AccountPlanType) -> bool {
diff --git a/codex-rs/core/src/session/session.rs b/codex-rs/core/src/session/session.rs
index 50b3345d61..73ba043c3d 100644
--- a/codex-rs/core/src/session/session.rs
+++ b/codex-rs/core/src/session/session.rs
@@ -132,7 +132,7 @@ impl SessionConfiguration {
         ThreadConfigSnapshot {
             model: self.collaboration_mode.model().to_string(),
             model_provider_id: self.original_config_do_not_use.model_provider_id.clone(),
-            service_tier: self.service_tier,
+            service_tier: self.service_tier.clone(),
             approval_policy: self.approval_policy.value(),
             approvals_reviewer: self.approvals_reviewer,
             permission_profile: self.permission_profile(),
@@ -177,7 +177,7 @@ impl SessionConfiguration {
         if let Some(summary) = updates.reasoning_summary {
             next_configuration.model_reasoning_summary = Some(summary);
         }
-        if let Some(service_tier) = updates.service_tier {
+        if let Some(service_tier) = updates.service_tier.clone() {
             next_configuration.service_tier = service_tier;
         }
         if let Some(personality) = updates.personality {
diff --git a/codex-rs/core/src/session/tests.rs b/codex-rs/core/src/session/tests.rs
index 89633daf35..4b7fa855ce 100644
--- a/codex-rs/core/src/session/tests.rs
+++ b/codex-rs/core/src/session/tests.rs
@@ -2801,7 +2801,7 @@ fn get_service_tier_defaults_enterprise_accounts_to_fast() {
             Some(AccountPlanType::Enterprise),
             /*fast_mode_enabled*/ true,
         ),
-        Some(ServiceTier::Fast)
+        Some(ServiceTier::priority())
     );
     assert_eq!(
         get_service_tier(
@@ -2810,7 +2810,7 @@ fn get_service_tier_defaults_enterprise_accounts_to_fast() {
             Some(AccountPlanType::EnterpriseCbpUsageBased),
             /*fast_mode_enabled*/ true,
         ),
-        Some(ServiceTier::Fast)
+        Some(ServiceTier::priority())
     );
     assert_eq!(
         get_service_tier(
@@ -2819,7 +2819,7 @@ fn get_service_tier_defaults_enterprise_accounts_to_fast() {
             Some(AccountPlanType::Business),
             /*fast_mode_enabled*/ true,
         ),
-        Some(ServiceTier::Fast)
+        Some(ServiceTier::priority())
     );
     assert_eq!(
         get_service_tier(
@@ -2828,7 +2828,7 @@ fn get_service_tier_defaults_enterprise_accounts_to_fast() {
             Some(AccountPlanType::Team),
             /*fast_mode_enabled*/ true,
         ),
-        Some(ServiceTier::Fast)
+        Some(ServiceTier::priority())
     );
     assert_eq!(
         get_service_tier(
@@ -2837,7 +2837,7 @@ fn get_service_tier_defaults_enterprise_accounts_to_fast() {
             Some(AccountPlanType::SelfServeBusinessUsageBased),
             /*fast_mode_enabled*/ true,
         ),
-        Some(ServiceTier::Fast)
+        Some(ServiceTier::priority())
     );
 }
 
diff --git a/codex-rs/core/src/session/turn.rs b/codex-rs/core/src/session/turn.rs
index 2b37372a33..f22aaab1d1 100644
--- a/codex-rs/core/src/session/turn.rs
+++ b/codex-rs/core/src/session/turn.rs
@@ -697,7 +697,7 @@ async fn track_turn_resolved_config_analytics(
             permission_profile_cwd: turn_context.cwd.to_path_buf(),
             reasoning_effort: turn_context.reasoning_effort,
             reasoning_summary: Some(turn_context.reasoning_summary),
-            service_tier: turn_context.config.service_tier,
+            service_tier: turn_context.config.service_tier.clone(),
             approval_policy: turn_context.approval_policy.value(),
             approvals_reviewer: turn_context.config.approvals_reviewer,
             sandbox_network_access: turn_context.network_sandbox_policy().is_enabled(),
@@ -1839,7 +1839,7 @@ async fn try_run_sampling_request(
             &turn_context.session_telemetry,
             turn_context.reasoning_effort,
             turn_context.reasoning_summary,
-            turn_context.config.service_tier,
+            turn_context.config.service_tier.clone(),
             turn_metadata_header,
             &inference_trace,
         )
diff --git a/codex-rs/core/src/session/turn_context.rs b/codex-rs/core/src/session/turn_context.rs
index a87d2ab178..e40f9f6563 100644
--- a/codex-rs/core/src/session/turn_context.rs
+++ b/codex-rs/core/src/session/turn_context.rs
@@ -397,7 +397,7 @@ impl Session {
         per_turn_config.model_reasoning_effort =
             session_configuration.collaboration_mode.reasoning_effort();
         per_turn_config.model_reasoning_summary = session_configuration.model_reasoning_summary;
-        per_turn_config.service_tier = session_configuration.service_tier;
+        per_turn_config.service_tier = session_configuration.service_tier.clone();
         per_turn_config.personality = session_configuration.personality;
         per_turn_config.approvals_reviewer = session_configuration.approvals_reviewer;
         per_turn_config.permissions.permission_profile =
diff --git a/codex-rs/core/tests/suite/agent_websocket.rs b/codex-rs/core/tests/suite/agent_websocket.rs
index 305346afac..f765c3c137 100644
--- a/codex-rs/core/tests/suite/agent_websocket.rs
+++ b/codex-rs/core/tests/suite/agent_websocket.rs
@@ -269,7 +269,7 @@ async fn websocket_v2_first_turn_uses_updated_fast_tier_after_startup_prewarm()
     assert_eq!(warmup["generate"].as_bool(), Some(false));
     assert_eq!(warmup.get("service_tier"), None);
 
-    test.submit_turn_with_service_tier("hello", Some(ServiceTier::Fast))
+    test.submit_turn_with_service_tier("hello", Some(ServiceTier::priority()))
         .await?;
 
     assert_eq!(server.handshakes().len(), 1);
@@ -313,7 +313,7 @@ async fn websocket_v2_first_turn_drops_fast_tier_after_startup_prewarm() -> Resu
             .features
             .enable(Feature::ResponsesWebsocketsV2)
             .expect("test config should allow feature update");
-        config.service_tier = Some(ServiceTier::Fast);
+        config.service_tier = Some(ServiceTier::priority());
     });
     let test = builder.build_with_websocket_server(&server).await?;
 
@@ -385,7 +385,7 @@ async fn websocket_v2_next_turn_uses_updated_service_tier() -> Result<()> {
     assert_eq!(warmup["generate"].as_bool(), Some(false));
     assert_eq!(warmup.get("service_tier"), None);
 
-    test.submit_turn_with_service_tier("first", Some(ServiceTier::Fast))
+    test.submit_turn_with_service_tier("first", Some(ServiceTier::priority()))
         .await?;
     test.submit_turn_with_service_tier("second", /*service_tier*/ None)
         .await?;
diff --git a/codex-rs/core/tests/suite/model_switching.rs b/codex-rs/core/tests/suite/model_switching.rs
index 43ec50746e..aa8d6d8ceb 100644
--- a/codex-rs/core/tests/suite/model_switching.rs
+++ b/codex-rs/core/tests/suite/model_switching.rs
@@ -105,7 +105,7 @@ fn test_model_info(
         used_fallback_model_metadata: false,
         supports_search_tool: false,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: None,
@@ -297,7 +297,7 @@ async fn service_tier_change_is_applied_on_next_http_turn() -> Result<()> {
 
     let test = test_codex().build(&server).await?;
 
-    test.submit_turn_with_service_tier("fast turn", Some(ServiceTier::Fast))
+    test.submit_turn_with_service_tier("fast turn", Some(ServiceTier::priority()))
         .await?;
     test.submit_turn_with_service_tier("standard turn", /*service_tier*/ None)
         .await?;
@@ -323,7 +323,7 @@ async fn flex_service_tier_is_applied_to_http_turn() -> Result<()> {
 
     let test = test_codex().build(&server).await?;
 
-    test.submit_turn_with_service_tier("flex turn", Some(ServiceTier::Flex))
+    test.submit_turn_with_service_tier("flex turn", Some(ServiceTier::flex()))
         .await?;
 
     let request = resp_mock.single_request();
@@ -817,7 +817,7 @@ async fn model_switch_to_smaller_model_updates_token_context_window() -> Result<
         used_fallback_model_metadata: false,
         supports_search_tool: false,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: None,
diff --git a/codex-rs/core/tests/suite/models_cache_ttl.rs b/codex-rs/core/tests/suite/models_cache_ttl.rs
index e2688afc97..207027bfdd 100644
--- a/codex-rs/core/tests/suite/models_cache_ttl.rs
+++ b/codex-rs/core/tests/suite/models_cache_ttl.rs
@@ -340,7 +340,7 @@ fn test_remote_model(slug: &str, priority: i32) -> ModelInfo {
         visibility: ModelVisibility::List,
         supported_in_api: true,
         priority,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: None,
diff --git a/codex-rs/core/tests/suite/personality.rs b/codex-rs/core/tests/suite/personality.rs
index dde6d2ca51..de43ad7869 100644
--- a/codex-rs/core/tests/suite/personality.rs
+++ b/codex-rs/core/tests/suite/personality.rs
@@ -578,7 +578,7 @@ async fn remote_model_friendly_personality_instructions_with_feature() -> anyhow
         visibility: ModelVisibility::List,
         supported_in_api: true,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: Some(ModelMessages {
@@ -686,7 +686,7 @@ async fn user_turn_personality_remote_model_template_includes_update_message() -
         visibility: ModelVisibility::List,
         supported_in_api: true,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: Some(ModelMessages {
diff --git a/codex-rs/core/tests/suite/remote_models.rs b/codex-rs/core/tests/suite/remote_models.rs
index 49218c78d2..4431d9e2c9 100644
--- a/codex-rs/core/tests/suite/remote_models.rs
+++ b/codex-rs/core/tests/suite/remote_models.rs
@@ -532,7 +532,7 @@ async fn remote_models_remote_model_uses_unified_exec() -> Result<()> {
         used_fallback_model_metadata: false,
         supports_search_tool: false,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: None,
@@ -788,7 +788,7 @@ async fn remote_models_apply_remote_base_instructions() -> Result<()> {
         used_fallback_model_metadata: false,
         supports_search_tool: false,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: remote_base.to_string(),
         model_messages: None,
@@ -1278,7 +1278,7 @@ fn test_remote_model_with_policy(
         used_fallback_model_metadata: false,
         supports_search_tool: false,
         priority,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: None,
diff --git a/codex-rs/core/tests/suite/rmcp_client.rs b/codex-rs/core/tests/suite/rmcp_client.rs
index 0947f4fba7..79469c20a5 100644
--- a/codex-rs/core/tests/suite/rmcp_client.rs
+++ b/codex-rs/core/tests/suite/rmcp_client.rs
@@ -1282,7 +1282,7 @@ async fn stdio_image_responses_are_sanitized_for_text_only_model() -> anyhow::Re
                 visibility: ModelVisibility::List,
                 supported_in_api: true,
                 priority: 1,
-                additional_speed_tiers: Vec::new(),
+                service_tiers: Vec::new(),
                 upgrade: None,
                 base_instructions: "base instructions".to_string(),
                 model_messages: None,
diff --git a/codex-rs/core/tests/suite/spawn_agent_description.rs b/codex-rs/core/tests/suite/spawn_agent_description.rs
index 031c3135e8..12845a0188 100644
--- a/codex-rs/core/tests/suite/spawn_agent_description.rs
+++ b/codex-rs/core/tests/suite/spawn_agent_description.rs
@@ -66,7 +66,7 @@ fn test_model_info(
         used_fallback_model_metadata: false,
         supports_search_tool: false,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: None,
diff --git a/codex-rs/core/tests/suite/view_image.rs b/codex-rs/core/tests/suite/view_image.rs
index 29c660d3af..a06f5e555f 100644
--- a/codex-rs/core/tests/suite/view_image.rs
+++ b/codex-rs/core/tests/suite/view_image.rs
@@ -1084,7 +1084,7 @@ async fn view_image_tool_returns_unsupported_message_for_text_only_model() -> an
         used_fallback_model_metadata: false,
         supports_search_tool: false,
         priority: 1,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         upgrade: None,
         base_instructions: "base instructions".to_string(),
         model_messages: None,
diff --git a/codex-rs/memories/write/src/startup_tests.rs b/codex-rs/memories/write/src/startup_tests.rs
index d89b68825e..1b1053fa44 100644
--- a/codex-rs/memories/write/src/startup_tests.rs
+++ b/codex-rs/memories/write/src/startup_tests.rs
@@ -253,14 +253,14 @@ async fn memories_startup_phase1_uses_live_thread_service_tier() -> anyhow::Resu
             model: None,
             effort: None,
             summary: None,
-            service_tier: Some(Some(ServiceTier::Fast)),
+            service_tier: Some(Some(ServiceTier::priority())),
             collaboration_mode: None,
             personality: None,
         })
         .await?;
 
-    let config_snapshot = wait_for_service_tier(&test, Some(ServiceTier::Fast)).await?;
-    assert_eq!(config_snapshot.service_tier, Some(ServiceTier::Fast));
+    let config_snapshot = wait_for_service_tier(&test, Some(ServiceTier::priority())).await?;
+    assert_eq!(config_snapshot.service_tier, Some(ServiceTier::priority()));
 
     let context = crate::runtime::MemoryStartupContext::new(
         Arc::clone(&test.thread_manager),
@@ -277,7 +277,7 @@ async fn memories_startup_phase1_uses_live_thread_service_tier() -> anyhow::Resu
             ReasoningEffort::Low,
         )
         .await;
-    assert_eq!(request_context.service_tier, Some(ServiceTier::Fast));
+    assert_eq!(request_context.service_tier, Some(ServiceTier::priority()));
 
     shutdown_test_codex(&test).await?;
     Ok(())
diff --git a/codex-rs/model-provider/src/amazon_bedrock/catalog.rs b/codex-rs/model-provider/src/amazon_bedrock/catalog.rs
index 4ca2cb891e..47d2c75619 100644
--- a/codex-rs/model-provider/src/amazon_bedrock/catalog.rs
+++ b/codex-rs/model-provider/src/amazon_bedrock/catalog.rs
@@ -1,10 +1,12 @@
 use codex_models_manager::model_info::BASE_INSTRUCTIONS;
 use codex_protocol::config_types::ReasoningSummary;
+use codex_protocol::config_types::ServiceTier;
 use codex_protocol::config_types::Verbosity;
 use codex_protocol::openai_models::ApplyPatchToolType;
 use codex_protocol::openai_models::ConfigShellToolType;
 use codex_protocol::openai_models::InputModality;
 use codex_protocol::openai_models::ModelInfo;
+use codex_protocol::openai_models::ModelServiceTier;
 use codex_protocol::openai_models::ModelVisibility;
 use codex_protocol::openai_models::ModelsResponse;
 use codex_protocol::openai_models::ReasoningEffort;
@@ -46,7 +48,11 @@ fn gpt_5_4_cmb_bedrock_model(priority: i32) -> ModelInfo {
         visibility: ModelVisibility::List,
         supported_in_api: true,
         priority,
-        additional_speed_tiers: vec!["fast".to_string()],
+        service_tiers: vec![ModelServiceTier {
+            id: ServiceTier::priority(),
+            name: "Fast".to_string(),
+            description: "Fast tier".to_string(),
+        }],
         availability_nux: None,
         upgrade: None,
         base_instructions: BASE_INSTRUCTIONS.to_string(),
@@ -86,7 +92,7 @@ fn bedrock_oss_model(slug: &str, display_name: &str, priority: i32) -> ModelInfo
         visibility: ModelVisibility::List,
         supported_in_api: true,
         priority,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         availability_nux: None,
         upgrade: None,
         base_instructions: BASE_INSTRUCTIONS.to_string(),
diff --git a/codex-rs/models-manager/models.json b/codex-rs/models-manager/models.json
index c025fe36ad..ce06a3dd35 100644
--- a/codex-rs/models-manager/models.json
+++ b/codex-rs/models-manager/models.json
@@ -82,8 +82,12 @@
         "team"
       ],
       "supports_search_tool": true,
-      "additional_speed_tiers": [
-        "fast"
+      "service_tiers": [
+        {
+          "id": "priority",
+          "name": "Fast",
+          "description": "Fast tier"
+        }
       ],
       "supports_reasoning_summaries": true
     },
@@ -164,8 +168,12 @@
         "team"
       ],
       "supports_search_tool": true,
-      "additional_speed_tiers": [
-        "fast"
+      "service_tiers": [
+        {
+          "id": "priority",
+          "name": "Fast",
+          "description": "Fast tier"
+        }
       ],
       "supports_reasoning_summaries": true
     },
@@ -249,7 +257,7 @@
         "team"
       ],
       "supports_search_tool": true,
-      "additional_speed_tiers": [],
+      "service_tiers": [],
       "supports_reasoning_summaries": true
     },
     {
@@ -332,7 +340,7 @@
         "team"
       ],
       "supports_search_tool": true,
-      "additional_speed_tiers": [],
+      "service_tiers": [],
       "supports_reasoning_summaries": true
     },
     {
@@ -411,7 +419,7 @@
         "team"
       ],
       "supports_search_tool": true,
-      "additional_speed_tiers": [],
+      "service_tiers": [],
       "supports_reasoning_summaries": true
     },
     {
@@ -491,7 +499,7 @@
         "team"
       ],
       "supports_search_tool": true,
-      "additional_speed_tiers": [],
+      "service_tiers": [],
       "supports_reasoning_summaries": true
     }
   ]
diff --git a/codex-rs/models-manager/src/model_info.rs b/codex-rs/models-manager/src/model_info.rs
index 8e8abae549..5afcaf7ce2 100644
--- a/codex-rs/models-manager/src/model_info.rs
+++ b/codex-rs/models-manager/src/model_info.rs
@@ -75,7 +75,7 @@ pub fn model_info_from_slug(slug: &str) -> ModelInfo {
         visibility: ModelVisibility::None,
         supported_in_api: true,
         priority: 99,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         availability_nux: None,
         upgrade: None,
         base_instructions: BASE_INSTRUCTIONS.to_string(),
diff --git a/codex-rs/protocol/src/config_types.rs b/codex-rs/protocol/src/config_types.rs
index da83ee858a..b4433aeae2 100644
--- a/codex-rs/protocol/src/config_types.rs
+++ b/codex-rs/protocol/src/config_types.rs
@@ -347,12 +347,88 @@ impl From<WebSearchToolConfig> for WebSearchConfig {
     }
 }
 
-#[derive(Debug, Serialize, Deserialize, Clone, Copy, PartialEq, Eq, Display, JsonSchema, TS)]
-#[serde(rename_all = "lowercase")]
-#[strum(serialize_all = "lowercase")]
-pub enum ServiceTier {
-    Fast,
-    Flex,
+pub const SERVICE_TIER_PRIORITY: &str = "priority";
+pub const SERVICE_TIER_FLEX: &str = "flex";
+
+#[derive(Debug, Serialize, Deserialize, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, TS)]
+#[serde(transparent)]
+#[ts(type = "string")]
+pub struct ServiceTier(String);
+
+impl ServiceTier {
+    pub fn new(id: impl Into<String>) -> Self {
+        Self(id.into())
+    }
+
+    pub fn as_str(&self) -> &str {
+        self.0.as_str()
+    }
+
+    pub fn priority() -> Self {
+        Self::new(SERVICE_TIER_PRIORITY)
+    }
+
+    pub fn flex() -> Self {
+        Self::new(SERVICE_TIER_FLEX)
+    }
+
+    pub fn is_priority(&self) -> bool {
+        self.as_str() == SERVICE_TIER_PRIORITY
+    }
+
+    pub fn is_flex(&self) -> bool {
+        self.as_str() == SERVICE_TIER_FLEX
+    }
+}
+
+impl AsRef<str> for ServiceTier {
+    fn as_ref(&self) -> &str {
+        self.as_str()
+    }
+}
+
+impl From<&str> for ServiceTier {
+    fn from(value: &str) -> Self {
+        Self::new(value)
+    }
+}
+
+impl From<String> for ServiceTier {
+    fn from(value: String) -> Self {
+        Self::new(value)
+    }
+}
+
+impl From<ServiceTier> for String {
+    fn from(value: ServiceTier) -> Self {
+        value.0
+    }
+}
+
+impl std::fmt::Display for ServiceTier {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.write_str(self.as_str())
+    }
+}
+
+impl JsonSchema for ServiceTier {
+    fn schema_name() -> String {
+        "ServiceTier".to_string()
+    }
+
+    fn json_schema(_generator: &mut SchemaGenerator) -> Schema {
+        Schema::Object(SchemaObject {
+            instance_type: Some(InstanceType::String.into()),
+            metadata: Some(Box::new(Metadata {
+                description: Some(
+                    "String-backed service tier identifier. Known values include `priority` and `flex`, but other backend-provided ids are allowed."
+                        .to_string(),
+                ),
+                ..Default::default()
+            })),
+            ..Default::default()
+        })
+    }
 }
 
 #[derive(Debug, Serialize, Deserialize, Clone, Copy, PartialEq, Eq, Display, JsonSchema, TS)]
diff --git a/codex-rs/protocol/src/openai_models.rs b/codex-rs/protocol/src/openai_models.rs
index 41275e6a6b..6f2daccea9 100644
--- a/codex-rs/protocol/src/openai_models.rs
+++ b/codex-rs/protocol/src/openai_models.rs
@@ -17,10 +17,10 @@ use ts_rs::TS;
 
 use crate::config_types::Personality;
 use crate::config_types::ReasoningSummary;
+use crate::config_types::ServiceTier;
 use crate::config_types::Verbosity;
 
 const PERSONALITY_PLACEHOLDER: &str = "{{ personality }}";
-pub const SPEED_TIER_FAST: &str = "fast";
 
 /// See https://platform.openai.com/docs/guides/reasoning?api-mode=responses#get-started-with-reasoning
 #[derive(
@@ -115,6 +115,13 @@ pub struct ModelAvailabilityNux {
     pub message: String,
 }
 
+#[derive(Debug, Clone, Deserialize, Serialize, TS, JsonSchema, PartialEq, Eq)]
+pub struct ModelServiceTier {
+    pub id: ServiceTier,
+    pub name: String,
+    pub description: String,
+}
+
 /// Metadata describing a Codex-supported model.
 #[derive(Debug, Clone, Deserialize, Serialize, TS, JsonSchema, PartialEq)]
 pub struct ModelPreset {
@@ -133,9 +140,9 @@ pub struct ModelPreset {
     /// Whether this model supports personality-specific instructions.
     #[serde(default)]
     pub supports_personality: bool,
-    /// Additional speed tiers this model can run with beyond the standard path.
+    /// User-selectable service tiers supported by this model.
     #[serde(default)]
-    pub additional_speed_tiers: Vec<String>,
+    pub service_tiers: Vec<ModelServiceTier>,
     /// Whether this is the default model for new users.
     pub is_default: bool,
     /// recommended upgrade model
@@ -257,7 +264,7 @@ pub struct ModelInfo {
     pub supported_in_api: bool,
     pub priority: i32,
     #[serde(default)]
-    pub additional_speed_tiers: Vec<String>,
+    pub service_tiers: Vec<ModelServiceTier>,
     pub availability_nux: Option<ModelAvailabilityNux>,
     pub upgrade: Option<ModelInfoUpgrade>,
     pub base_instructions: String,
@@ -441,7 +448,7 @@ impl From<ModelInfo> for ModelPreset {
                 .unwrap_or(ReasoningEffort::None),
             supported_reasoning_efforts: info.supported_reasoning_levels.clone(),
             supports_personality,
-            additional_speed_tiers: info.additional_speed_tiers,
+            service_tiers: info.service_tiers,
             is_default: false, // default is the highest priority available model
             upgrade: info.upgrade.as_ref().map(|upgrade| ModelUpgrade {
                 id: upgrade.model.clone(),
@@ -463,10 +470,16 @@ impl From<ModelInfo> for ModelPreset {
 }
 
 impl ModelPreset {
-    pub fn supports_fast_mode(&self) -> bool {
-        self.additional_speed_tiers
+    pub fn supports_service_tier(&self, service_tier: &ServiceTier) -> bool {
+        self.service_tiers
             .iter()
-            .any(|tier| tier == SPEED_TIER_FAST)
+            .any(|tier| tier.id == *service_tier)
+    }
+
+    pub fn service_tier(&self, service_tier: &ServiceTier) -> Option<&ModelServiceTier> {
+        self.service_tiers
+            .iter()
+            .find(|tier| tier.id == *service_tier)
     }
 
     /// Filter models based on authentication mode.
@@ -547,7 +560,7 @@ mod tests {
             visibility: ModelVisibility::List,
             supported_in_api: true,
             priority: 1,
-            additional_speed_tiers: Vec::new(),
+            service_tiers: Vec::new(),
             availability_nux: None,
             upgrade: None,
             base_instructions: "base".to_string(),
@@ -817,7 +830,11 @@ mod tests {
             availability_nux: Some(ModelAvailabilityNux {
                 message: "Try Spark.".to_string(),
             }),
-            additional_speed_tiers: vec![SPEED_TIER_FAST.to_string()],
+            service_tiers: vec![ModelServiceTier {
+                id: ServiceTier::priority(),
+                name: "Fast".to_string(),
+                description: "Fast tier".to_string(),
+            }],
             ..test_model(/*spec*/ None)
         });
 
@@ -827,6 +844,6 @@ mod tests {
                 message: "Try Spark.".to_string(),
             })
         );
-        assert!(preset.supports_fast_mode());
+        assert!(preset.supports_service_tier(&ServiceTier::priority()));
     }
 }
diff --git a/codex-rs/tools/src/agent_tool_tests.rs b/codex-rs/tools/src/agent_tool_tests.rs
index 3157cfc547..46fed4db23 100644
--- a/codex-rs/tools/src/agent_tool_tests.rs
+++ b/codex-rs/tools/src/agent_tool_tests.rs
@@ -19,7 +19,7 @@ fn model_preset(id: &str, show_in_picker: bool) -> ModelPreset {
             description: "Balanced".to_string(),
         }],
         supports_personality: false,
-        additional_speed_tiers: Vec::new(),
+        service_tiers: Vec::new(),
         is_default: false,
         upgrade: None,
         show_in_picker,