feat: Added full cost tracking to Coyote and added it to the default right prompt

src/client/common.rs

@@ -402,6 +402,27 @@ impl TokenUsage {
             other.cache_read_input_tokens,
         );
     }
+
+    /// USD cost of this usage at the given model's prices (USD per 1M
+    /// tokens). The four buckets are disjoint on both Anthropic direct and
+    /// Bedrock Converse (`input_tokens` excludes cache tokens), so the terms
+    /// sum without adjustment. Strict on missing prices: any nonzero
+    /// component without a price yields `None` rather than a silent
+    /// undercount. All-zero usage is also `None` — nothing billed, nothing
+    /// to show.
+    pub fn cost_usd(&self, model: &Model) -> Option<f64> {
+        fn term(tokens: Option<u64>, price: Option<f64>) -> Option<f64> {
+            match tokens.unwrap_or(0) {
+                0 => Some(0.0),
+                tokens => Some(tokens as f64 * price? / 1_000_000.0),
+            }
+        }
+        let total = term(self.input_tokens, model.input_price())?
+            + term(self.output_tokens, model.output_price())?
+            + term(self.cache_read_input_tokens, model.cache_read_price())?
+            + term(self.cache_creation_input_tokens, model.cache_write_price())?;
+        (total > 0.0).then_some(total)
+    }
 }
 
 #[derive(Debug)]
@@ -559,7 +580,7 @@ pub async fn call_chat_completions(
                     ctx.app.config.print_markdown(&text)?;
                 }
             }
-            ctx.record_token_usage(usage);
+            ctx.record_token_usage(usage, client.model());
             finish_completion(ctx, text, tool_calls, thinking).await
         }
         Err(err) => Err(err),
@@ -591,7 +612,7 @@ pub async fn call_chat_completions_streaming(
     // call's input tokens (reported in message_start), so abort and error
     // paths must still count toward the session totals.
     let (text, tool_calls, thinking, usage) = handler.take();
-    ctx.record_token_usage(usage);
+    ctx.record_token_usage(usage, client.model());
 
     if aborted_ctrld {
         bail!("Aborted.");
@@ -665,7 +686,7 @@ pub async fn call_chat_completions_streaming_quiet(
             thinking,
             usage,
         } = output;
-        ctx.record_token_usage(usage);
+        ctx.record_token_usage(usage, client.model());
         if abort_signal.aborted() {
             bail!("Aborted.");
         }
@@ -681,7 +702,7 @@ pub async fn call_chat_completions_streaming_quiet(
     let send_ret = client.chat_completions_streaming(input, &mut handler).await;
 
     let (text, tool_calls, thinking, usage) = handler.take();
-    ctx.record_token_usage(usage);
+    ctx.record_token_usage(usage, client.model());
 
     if abort_signal.aborted() {
         bail!("Aborted.");
@@ -1000,6 +1021,72 @@ mod tests {
         assert_eq!(total.input_tokens, Some(u64::MAX));
     }
 
+    fn priced_model(
+        input_price: Option<f64>,
+        output_price: Option<f64>,
+        cache_read_price: Option<f64>,
+        cache_write_price: Option<f64>,
+    ) -> Model {
+        let mut data = ModelData::new("test");
+        data.input_price = input_price;
+        data.output_price = output_price;
+        data.cache_read_price = cache_read_price;
+        data.cache_write_price = cache_write_price;
+        Model::from_config("provider", &[data]).remove(0)
+    }
+
+    #[test]
+    fn test_token_usage_cost_usd_sums_all_four_components() {
+        let usage = TokenUsage {
+            input_tokens: Some(1_000_000),
+            output_tokens: Some(500_000),
+            cache_creation_input_tokens: Some(200_000),
+            cache_read_input_tokens: Some(2_000_000),
+        };
+        let model = priced_model(Some(3.0), Some(15.0), Some(0.3), Some(3.75));
+        // 3.0 + 7.5 + 0.6 + 0.75
+        let cost = usage.cost_usd(&model).unwrap();
+        assert!((cost - 11.85).abs() < 1e-9);
+    }
+
+    #[test]
+    fn test_token_usage_cost_usd_is_none_when_a_billed_component_lacks_a_price() {
+        let usage = TokenUsage {
+            input_tokens: Some(100),
+            output_tokens: Some(10),
+            cache_creation_input_tokens: None,
+            cache_read_input_tokens: Some(50),
+        };
+        let model = priced_model(Some(3.0), Some(15.0), None, Some(3.75));
+        assert_eq!(usage.cost_usd(&model), None);
+    }
+
+    #[test]
+    fn test_token_usage_cost_usd_ignores_missing_prices_for_unbilled_components() {
+        let usage = TokenUsage {
+            input_tokens: Some(1_000_000),
+            output_tokens: Some(1_000_000),
+            cache_creation_input_tokens: Some(0),
+            cache_read_input_tokens: None,
+        };
+        let model = priced_model(Some(3.0), Some(15.0), None, None);
+        let cost = usage.cost_usd(&model).unwrap();
+        assert!((cost - 18.0).abs() < 1e-9);
+    }
+
+    #[test]
+    fn test_token_usage_cost_usd_is_none_for_empty_or_all_zero_usage() {
+        let model = priced_model(Some(3.0), Some(15.0), Some(0.3), Some(3.75));
+        assert_eq!(TokenUsage::default().cost_usd(&model), None);
+        let zero = TokenUsage {
+            input_tokens: Some(0),
+            output_tokens: Some(0),
+            cache_creation_input_tokens: Some(0),
+            cache_read_input_tokens: Some(0),
+        };
+        assert_eq!(zero.cost_usd(&model), None);
+    }
+
     #[test]
     fn test_catch_error_display_json_with_type() {
         let data = json!({"error": {"type": "invalid_request_error", "message": "Bad request"}});

src/client/model.rs

@@ -195,6 +195,22 @@ impl Model {
         self.data.max_output_tokens
     }
 
+    pub fn input_price(&self) -> Option<f64> {
+        self.data.input_price
+    }
+
+    pub fn output_price(&self) -> Option<f64> {
+        self.data.output_price
+    }
+
+    pub fn cache_read_price(&self) -> Option<f64> {
+        self.data.cache_read_price
+    }
+
+    pub fn cache_write_price(&self) -> Option<f64> {
+        self.data.cache_write_price
+    }
+
     pub fn supports_function_calling(&self) -> bool {
         self.data.supports_function_calling
     }

src/config/mod.rs

@@ -218,7 +218,7 @@ const SUMMARY_CONTEXT_PROMPT: &str = "This is a summary of the chat history as a
 
 const LEFT_PROMPT: &str = "{color.red}{model}){color.green}{?session {?agent {agent}>}{session}{?role /}}{!session {?agent {agent}>}}{role}{?rag @{rag}}{color.cyan}{?session )}{!session >}{color.reset} ";
 // `{?var}` treats "0" as falsy, so a reported-but-zero cache read intentionally hides the ⚡ segment.
-const RIGHT_PROMPT: &str = "{color.cyan}{?reasoning_effort [{reasoning_effort}] }{color.purple}{?last_cache_read_tokens ⚡{last_cache_read_tokens} }{?session {?consume_tokens {consume_tokens}({consume_percent}%)}{!consume_tokens {consume_tokens}}}{color.reset}";
+const RIGHT_PROMPT: &str = "{color.cyan}{?reasoning_effort [{reasoning_effort}] }{color.purple}{?last_cache_read_tokens ⚡{last_cache_read_tokens} }{?total_cost ${total_cost} }{?session {?consume_tokens {consume_tokens}({consume_percent}%)}{!consume_tokens {consume_tokens}}}{color.reset}";
 
 static EDITOR: OnceLock<Option<String>> = OnceLock::new();
 

src/config/request_context.rs

@@ -311,6 +311,7 @@ pub struct RequestContext {
     /// Usage reported by the most recent API call; reset on every completion
     /// so providers that report nothing never show stale numbers.
     pub last_token_usage: Option<TokenUsage>,
+    pub last_cost: Option<f64>,
 
     pub tool_scope: ToolScope,
 
@@ -372,6 +373,7 @@ impl RequestContext {
             agent: None,
             last_message: None,
             last_token_usage: None,
+            last_cost: None,
             tool_scope: ToolScope::default(),
             declared_function_names: Default::default(),
             node_job_scope: None,
@@ -437,6 +439,7 @@ impl RequestContext {
             agent: None,
             last_message: None,
             last_token_usage: None,
+            last_cost: None,
             tool_scope: ToolScope {
                 functions,
                 mcp_runtime,
@@ -499,6 +502,7 @@ impl RequestContext {
             agent: self.agent.clone(),
             last_message: self.last_message.clone(),
             last_token_usage: self.last_token_usage.clone(),
+            last_cost: self.last_cost,
             tool_scope: self.tool_scope.clone(),
             declared_function_names: self.declared_function_names.clone(),
             node_job_scope: None,
@@ -549,6 +553,7 @@ impl RequestContext {
             agent: None,
             last_message: None,
             last_token_usage: None,
+            last_cost: None,
             tool_scope: ToolScope {
                 functions: Functions::default(),
                 mcp_runtime: McpRuntime::default(),
@@ -2236,14 +2241,19 @@ impl RequestContext {
         }
     }
 
-    pub fn record_token_usage(&mut self, usage: Option<TokenUsage>) {
+    pub fn record_token_usage(&mut self, usage: Option<TokenUsage>, model: &Model) {
+        let cost = usage.as_ref().and_then(|u| u.cost_usd(model));
         if let Some(usage) = &usage {
             debug!("token-usage: {usage:?}");
             if let Some(session) = self.session.as_mut() {
                 session.accumulate_token_usage(usage);
+                if let Some(cost) = cost {
+                    session.accumulate_cost(cost);
+                }
             }
         }
         self.last_token_usage = usage;
+        self.last_cost = cost;
     }
 
     pub fn generate_prompt_context(&self, app: &AppConfig) -> HashMap<&str, String> {
@@ -2311,6 +2321,12 @@ impl RequestContext {
             if let Some(value) = usage.cache_creation_input_tokens {
                 output.insert("last_cache_creation_tokens", value.to_string());
             }
+            if let Some(value) = self.last_cost {
+                let rendered = format!("{value:.4}");
+                if rendered != "0.0000" {
+                    output.insert("last_cost", rendered);
+                }
+            }
         }
         if let Some(session) = &self.session {
             output.insert("session", session.name().to_string());
@@ -2335,6 +2351,10 @@ impl RequestContext {
             if let Some(value) = usage.cache_creation_input_tokens {
                 output.insert("total_cache_creation_tokens", value.to_string());
             }
+            let rendered = format!("{:.4}", session.cost());
+            if rendered != "0.0000" {
+                output.insert("total_cost", rendered);
+            }
         }
         if let Some(rag) = &self.rag {
             output.insert("rag", rag.name().to_string());
@@ -5343,6 +5363,7 @@ fn fork_base_name(name: &str) -> &str {
 mod tests {
     use super::super::mcp_factory::McpFactory;
     use super::*;
+    use crate::client::ModelData;
     use crate::config::AppState;
     use crate::config::agent::AgentConfig;
     use crate::config::bundles::BundleStore;
@@ -5460,22 +5481,37 @@ mod tests {
         RequestContext::new(default_app_state(), WorkingMode::Cmd)
     }
 
+    fn priced_model() -> Model {
+        let mut data = ModelData::new("test");
+        data.input_price = Some(10.0);
+        data.output_price = Some(100.0);
+        data.cache_read_price = Some(1.0);
+        data.cache_write_price = Some(20.0);
+        Model::from_config("provider", &[data]).remove(0)
+    }
+
     #[test]
     fn generate_prompt_context_emits_token_usage_vars() {
         let mut ctx = create_test_ctx();
         ctx.session = Some(Session::default());
-        ctx.record_token_usage(Some(TokenUsage {
-            input_tokens: Some(120),
-            output_tokens: Some(8),
-            cache_creation_input_tokens: None,
-            cache_read_input_tokens: Some(100),
-        }));
-        ctx.record_token_usage(Some(TokenUsage {
-            input_tokens: Some(30),
-            output_tokens: Some(2),
-            cache_creation_input_tokens: Some(40),
-            cache_read_input_tokens: None,
-        }));
+        ctx.record_token_usage(
+            Some(TokenUsage {
+                input_tokens: Some(120),
+                output_tokens: Some(8),
+                cache_creation_input_tokens: None,
+                cache_read_input_tokens: Some(100),
+            }),
+            &priced_model(),
+        );
+        ctx.record_token_usage(
+            Some(TokenUsage {
+                input_tokens: Some(30),
+                output_tokens: Some(2),
+                cache_creation_input_tokens: Some(40),
+                cache_read_input_tokens: None,
+            }),
+            &priced_model(),
+        );
 
         let app = ctx.app.config.clone();
         let vars = ctx.generate_prompt_context(&app);
@@ -5509,21 +5545,49 @@ mod tests {
             vars.get("total_cache_creation_tokens").map(String::as_str),
             Some("40")
         );
+        // 30×10 + 2×100 + 40×20 per 1M = 0.0013; first call adds 0.0021.
+        assert_eq!(vars.get("last_cost").map(String::as_str), Some("0.0013"));
+        assert_eq!(vars.get("total_cost").map(String::as_str), Some("0.0034"));
     }
 
     #[test]
     fn generate_prompt_context_omits_token_usage_vars_when_provider_reports_none() {
         let mut ctx = create_test_ctx();
-        ctx.record_token_usage(Some(TokenUsage {
-            input_tokens: Some(1),
-            ..Default::default()
-        }));
-        ctx.record_token_usage(None);
+        ctx.record_token_usage(
+            Some(TokenUsage {
+                input_tokens: Some(1),
+                ..Default::default()
+            }),
+            &priced_model(),
+        );
+        ctx.record_token_usage(None, &priced_model());
 
         let app = ctx.app.config.clone();
         let vars = ctx.generate_prompt_context(&app);
         assert!(!vars.contains_key("last_input_tokens"));
         assert!(!vars.contains_key("total_input_tokens"));
+        assert!(!vars.contains_key("last_cost"));
+        assert!(!vars.contains_key("total_cost"));
+    }
+
+    #[test]
+    fn generate_prompt_context_omits_costs_that_render_as_zero() {
+        let mut ctx = create_test_ctx();
+        ctx.session = Some(Session::default());
+        // 1 input token at $10/1M is 0.00001, which formats to "0.0000".
+        ctx.record_token_usage(
+            Some(TokenUsage {
+                input_tokens: Some(1),
+                ..Default::default()
+            }),
+            &priced_model(),
+        );
+
+        let app = ctx.app.config.clone();
+        let vars = ctx.generate_prompt_context(&app);
+        assert_eq!(vars.get("last_input_tokens").map(String::as_str), Some("1"));
+        assert!(!vars.contains_key("last_cost"));
+        assert!(!vars.contains_key("total_cost"));
     }
 
     fn test_decl(name: &str) -> FunctionDeclaration {

src/config/session.rs

@@ -24,6 +24,10 @@ pub const INTERRUPTED_RESPONSE_TEXT: &str = "[Response interrupted due to error]
 /// instead of thrashing every turn.
 const COMPRESSION_MIN_RECLAIM_DIVISOR: usize = 5;
 
+fn cost_is_zero(v: &f64) -> bool {
+    *v == 0.0
+}
+
 #[derive(Debug, Clone, Default, Deserialize, Serialize)]
 pub struct Session {
     #[serde(rename(serialize = "model", deserialize = "model"))]
@@ -101,6 +105,8 @@ pub struct Session {
     auto_continue_paused: Option<String>,
     #[serde(default, skip_serializing_if = "TokenUsage::is_empty")]
     token_usage: TokenUsage,
+    #[serde(default, skip_serializing_if = "cost_is_zero")]
+    cost: f64,
 
     #[serde(skip)]
     model: Model,
@@ -276,6 +282,15 @@ impl Session {
         self.dirty = true;
     }
 
+    pub fn cost(&self) -> f64 {
+        self.cost
+    }
+
+    pub fn accumulate_cost(&mut self, cost: f64) {
+        self.cost += cost;
+        self.dirty = true;
+    }
+
     pub fn has_user_messages(&self) -> bool {
         self.messages.iter().any(|v| v.role.is_user())
     }
@@ -357,6 +372,9 @@ impl Session {
         if let Some(value) = self.token_usage.cache_creation_input_tokens {
             data["cache_creation_tokens"] = value.into();
         }
+        if self.cost != 0.0 {
+            data["total_cost"] = ((self.cost * 1e6).round() / 1e6).into();
+        }
         data["messages"] = json!(self.messages);
 
         let output = serde_yaml::to_string(&data)
@@ -467,6 +485,9 @@ impl Session {
         if let Some(value) = self.token_usage.cache_creation_input_tokens {
             items.push(("cache_creation_tokens", value.to_string()));
         }
+        if self.cost != 0.0 {
+            items.push(("total_cost", format!("${:.4}", self.cost)));
+        }
 
         let mut lines: Vec<String> = items
             .iter()
@@ -1212,6 +1233,28 @@ mod tests {
         assert!(!yaml.contains("token_usage"));
     }
 
+    #[test]
+    fn session_cost_accumulates_and_survives_yaml_round_trip() {
+        let mut session = Session::default();
+        assert_eq!(session.cost(), 0.0);
+
+        session.accumulate_cost(0.0025);
+        session.accumulate_cost(0.001);
+        assert!((session.cost() - 0.0035).abs() < 1e-12);
+        assert!(session.dirty());
+
+        let yaml = serde_yaml::to_string(&session).unwrap();
+        let reloaded: Session = serde_yaml::from_str(&yaml).unwrap();
+        assert!((reloaded.cost() - 0.0035).abs() < 1e-12);
+    }
+
+    #[test]
+    fn session_zero_cost_is_not_serialized() {
+        let session = Session::default();
+        let yaml = serde_yaml::to_string(&session).unwrap();
+        assert!(!yaml.contains("cost"));
+    }
+
     #[test]
     fn session_export_includes_real_usage_only_when_reported() {
         let empty = Session::default().export().unwrap();