fix

codex-rs/core/src/codex.rs

@@ -830,7 +830,7 @@ impl Session {
 
     async fn get_total_token_usage(&self) -> i64 {
         let state = self.state.lock().await;
-        state.get_total_token_usage(state.server_reasoning_included())
+        state.get_total_token_usage()
     }
 
     pub(crate) async fn get_base_instructions(&self) -> BaseInstructions {

codex-rs/core/src/context_manager/history.rs

@@ -25,6 +25,7 @@ pub(crate) struct ContextManager {
     /// The oldest items are at the beginning of the vector.
     items: Vec<ResponseItem>,
     token_info: Option<TokenUsageInfo>,
+    server_reasoning_included: bool,
     codex_home: PathBuf,
 }
 
@@ -33,6 +34,7 @@ impl ContextManager {
         Self {
             items: Vec::new(),
             token_info: TokenUsageInfo::new_or_append(&None, &None, None),
+            server_reasoning_included: false,
             codex_home: codex_home.to_path_buf(),
         }
     }
@@ -54,6 +56,14 @@ impl ContextManager {
         }
     }
 
+    pub(crate) fn set_server_reasoning_included(&mut self, included: bool) {
+        self.server_reasoning_included = included;
+    }
+
+    pub(crate) fn total_token_usage(&self) -> i64 {
+        self.get_total_token_usage(self.server_reasoning_included)
+    }
+
     /// `items` is ordered from oldest to newest.
     pub(crate) fn record_items<I>(&mut self, items: I, policy: TruncationPolicy)
     where
@@ -321,7 +331,8 @@ impl ContextManager {
             return item.clone();
         };
 
-        let mut remaining = context_window.saturating_sub(self.get_total_token_usage());
+        let mut remaining = context_window
+            .saturating_sub(self.get_total_token_usage(self.server_reasoning_included));
         let offloader = ContextOffloader::new(&self.codex_home);
         let mut new_content = Vec::with_capacity(content.len().saturating_add(1));
         for item in content {

codex-rs/core/src/state/session.rs

@@ -14,7 +14,6 @@ pub(crate) struct SessionState {
     pub(crate) session_configuration: SessionConfiguration,
     pub(crate) history: ContextManager,
     pub(crate) latest_rate_limits: Option<RateLimitSnapshot>,
-    pub(crate) server_reasoning_included: bool,
 }
 
 impl SessionState {
@@ -25,7 +24,6 @@ impl SessionState {
             session_configuration,
             history,
             latest_rate_limits: None,
-            server_reasoning_included: false,
         }
     }
 
@@ -80,17 +78,12 @@ impl SessionState {
         self.history.set_token_usage_full(context_window);
     }
 
-    pub(crate) fn get_total_token_usage(&self, server_reasoning_included: bool) -> i64 {
-        self.history
-            .get_total_token_usage(server_reasoning_included)
+    pub(crate) fn get_total_token_usage(&self) -> i64 {
+        self.history.total_token_usage()
     }
 
     pub(crate) fn set_server_reasoning_included(&mut self, included: bool) {
-        self.server_reasoning_included = included;
-    }
-
-    pub(crate) fn server_reasoning_included(&self) -> bool {
-        self.server_reasoning_included
+        self.history.set_server_reasoning_included(included);
     }
 }