Correctly calculate remaining context size (#3190)

We had multiple issues with context size calculation:
1. `initial_prompt_tokens` calculation based on cache size is not
reliable, cache misses might set it to much higher value. For now
hardcoded to a safer constant.
2. Input context size for GPT-5 is 272k (that's where 33% came from).

Fixes.

codex-rs/core/src/config.rs

@@ -1382,7 +1382,7 @@ model_verbosity = "high"
         let expected_gpt5_profile_config = Config {
             model: "gpt-5".to_string(),
             model_family: find_family_for_model("gpt-5").expect("known model slug"),
-            model_context_window: Some(400_000),
+            model_context_window: Some(272_000),
             model_max_output_tokens: Some(128_000),
             model_provider_id: "openai".to_string(),
             model_provider: fixture.openai_provider.clone(),

codex-rs/core/src/openai_model_info.rs

@@ -79,12 +79,12 @@ pub(crate) fn get_model_info(model_family: &ModelFamily) -> Option<ModelInfo> {
         }),
 
         "gpt-5" => Some(ModelInfo {
-            context_window: 400_000,
+            context_window: 272_000,
             max_output_tokens: 128_000,
         }),
 
         _ if slug.starts_with("codex-") => Some(ModelInfo {
-            context_window: 400_000,
+            context_window: 272_000,
             max_output_tokens: 128_000,
         }),
 

codex-rs/protocol/src/protocol.rs

@@ -527,6 +527,9 @@ pub struct TokenUsage {
     pub total_tokens: u64,
 }
 
+// Includes prompts, tools and space to call compact.
+const BASELINE_TOKENS: u64 = 12000;
+
 impl TokenUsage {
     pub fn is_zero(&self) -> bool {
         self.total_tokens == 0
@@ -557,26 +560,22 @@ impl TokenUsage {
     /// Estimate the remaining user-controllable percentage of the model's context window.
     ///
     /// `context_window` is the total size of the model's context window.
-    /// `baseline_used_tokens` should capture tokens that are always present in
+    /// `BASELINE_TOKENS` should capture tokens that are always present in
     /// the context (e.g., system prompt and fixed tool instructions) so that
     /// the percentage reflects the portion the user can influence.
     ///
     /// This normalizes both the numerator and denominator by subtracting the
     /// baseline, so immediately after the first prompt the UI shows 100% left
     /// and trends toward 0% as the user fills the effective window.
-    pub fn percent_of_context_window_remaining(
-        &self,
-        context_window: u64,
-        baseline_used_tokens: u64,
-    ) -> u8 {
-        if context_window <= baseline_used_tokens {
+    pub fn percent_of_context_window_remaining(&self, context_window: u64) -> u8 {
+        if context_window <= BASELINE_TOKENS {
             return 0;
         }
 
-        let effective_window = context_window - baseline_used_tokens;
+        let effective_window = context_window - BASELINE_TOKENS;
         let used = self
             .tokens_in_context_window()
-            .saturating_sub(baseline_used_tokens);
+            .saturating_sub(BASELINE_TOKENS);
         let remaining = effective_window.saturating_sub(used);
         ((remaining as f32 / effective_window as f32) * 100.0).clamp(0.0, 100.0) as u8
     }

codex-rs/tui/src/bottom_pane/chat_composer.rs

@@ -67,15 +67,6 @@ struct TokenUsageInfo {
     total_token_usage: TokenUsage,
     last_token_usage: TokenUsage,
     model_context_window: Option<u64>,
-    /// Baseline token count present in the context before the user's first
-    /// message content is considered. This is used to normalize the
-    /// "context left" percentage so it reflects the portion the user can
-    /// influence rather than fixed prompt overhead (system prompt, tool
-    /// instructions, etc.).
-    ///
-    /// Preferred source is `cached_input_tokens` from the first turn (when
-    /// available), otherwise we fall back to 0.
-    initial_prompt_tokens: u64,
 }
 
 pub(crate) struct ChatComposer {
@@ -181,17 +172,10 @@ impl ChatComposer {
         last_token_usage: TokenUsage,
         model_context_window: Option<u64>,
     ) {
-        let initial_prompt_tokens = self
-            .token_usage_info
-            .as_ref()
-            .map(|info| info.initial_prompt_tokens)
-            .unwrap_or_else(|| last_token_usage.cached_input_tokens.unwrap_or(0));
-
         self.token_usage_info = Some(TokenUsageInfo {
             total_token_usage,
             last_token_usage,
             model_context_window,
-            initial_prompt_tokens,
         });
     }
 
@@ -1302,10 +1286,7 @@ impl WidgetRef for ChatComposer {
                     let last_token_usage = &token_usage_info.last_token_usage;
                     if let Some(context_window) = token_usage_info.model_context_window {
                         let percent_remaining: u8 = if context_window > 0 {
-                            last_token_usage.percent_of_context_window_remaining(
-                                context_window,
-                                token_usage_info.initial_prompt_tokens,
-                            )
+                            last_token_usage.percent_of_context_window_remaining(context_window)
                         } else {
                             100
                         };