From 3c2cf277bbd26625cf6240aa58e6e8cc52fbd0b8 Mon Sep 17 00:00:00 2001
From: Carsten Abele <carsten@ravensburg.dev>
Date: Tue, 7 Apr 2026 21:38:49 +0200
Subject: [PATCH] feat: add usage cost display in menu bar dropdown

Shows session and cumulative audio duration and cost ($0.006/min)
from the transcription.done event's usage.prompt_audio_seconds field.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 .../Models/TranscriptionManager.swift         | 16 ++++++++++++++-
 .../MyVoxtral/Network/VoxtralMessages.swift   | 19 ++++++++++++++++--
 MyVoxtral/MyVoxtral/Views/MenuBarView.swift   | 20 +++++++++++++++++++
 3 files changed, 52 insertions(+), 3 deletions(-)

diff --git a/MyVoxtral/MyVoxtral/Models/TranscriptionManager.swift b/MyVoxtral/MyVoxtral/Models/TranscriptionManager.swift
index 76f9312..0b483f9 100644
--- a/MyVoxtral/MyVoxtral/Models/TranscriptionManager.swift
+++ b/MyVoxtral/MyVoxtral/Models/TranscriptionManager.swift
@@ -8,8 +8,14 @@ enum RecordingState: Equatable {
 
 @MainActor
 final class TranscriptionManager: ObservableObject {
+    static let costPerSecond: Double = 0.0001 // $0.006/min for voxtral-mini-transcribe-realtime
+
     @Published var state: RecordingState = .idle
     @Published var currentText: String = ""
+    @Published var sessionAudioSeconds: Int = 0
+    @Published var totalAudioSeconds: Int = 0
+    @Published var sessionCost: Double = 0
+    @Published var totalCost: Double = 0
 
     private let audioCapture = AudioCapture()
     private let wsClient = VoxtralWebSocketClient()
@@ -33,6 +39,8 @@ final class TranscriptionManager: ObservableObject {
         }
 
         currentText = ""
+        sessionAudioSeconds = 0
+        sessionCost = 0
         hasRetried = false
 
         wsClient.onEvent = { [weak self] event in
@@ -79,10 +87,16 @@ final class TranscriptionManager: ObservableObject {
             break
         case .language:
             break
-        case .done(let text):
+        case .done(let text, let audioSeconds):
             if currentText.isEmpty {
                 currentText = text
             }
+            if let secs = audioSeconds {
+                sessionAudioSeconds = secs
+                sessionCost = Double(secs) * Self.costPerSecond
+                totalAudioSeconds += secs
+                totalCost += sessionCost
+            }
         case .error(let message):
             if !hasRetried && state == .recording {
                 hasRetried = true
diff --git a/MyVoxtral/MyVoxtral/Network/VoxtralMessages.swift b/MyVoxtral/MyVoxtral/Network/VoxtralMessages.swift
index 7f9edbb..832d154 100644
--- a/MyVoxtral/MyVoxtral/Network/VoxtralMessages.swift
+++ b/MyVoxtral/MyVoxtral/Network/VoxtralMessages.swift
@@ -47,7 +47,7 @@ enum VoxtralEvent {
     case textDelta(String)
     case segment(text: String, start: Double, end: Double)
     case language(String)
-    case done(text: String)
+    case done(text: String, audioSeconds: Int?)
     case error(String)
     case unknown(String)
 }
@@ -76,6 +76,21 @@ struct SegmentEvent: Decodable {
 
 struct DoneEvent: Decodable {
     let text: String
+    let usage: UsageInfo?
+}
+
+struct UsageInfo: Decodable {
+    let promptTokens: Int?
+    let completionTokens: Int?
+    let totalTokens: Int?
+    let promptAudioSeconds: Int?
+
+    enum CodingKeys: String, CodingKey {
+        case promptTokens = "prompt_tokens"
+        case completionTokens = "completion_tokens"
+        case totalTokens = "total_tokens"
+        case promptAudioSeconds = "prompt_audio_seconds"
+    }
 }
 
 struct ErrorEvent: Decodable {
@@ -111,7 +126,7 @@ func parseVoxtralEvent(from data: Data) -> VoxtralEvent {
         return .language(e.audioLanguage)
     case "transcription.done":
         guard let e = try? JSONDecoder().decode(DoneEvent.self, from: data) else { return .unknown("") }
-        return .done(text: e.text)
+        return .done(text: e.text, audioSeconds: e.usage?.promptAudioSeconds)
     case "error":
         if let e = try? JSONDecoder().decode(ErrorEvent.self, from: data) {
             return .error(e.error?.message?.detail ?? "Unknown error")
diff --git a/MyVoxtral/MyVoxtral/Views/MenuBarView.swift b/MyVoxtral/MyVoxtral/Views/MenuBarView.swift
index 0ce2ba0..ac56a71 100644
--- a/MyVoxtral/MyVoxtral/Views/MenuBarView.swift
+++ b/MyVoxtral/MyVoxtral/Views/MenuBarView.swift
@@ -20,6 +20,19 @@ struct MenuBarView: View {
                     .padding(.horizontal, 8)
             }
 
+            if manager.sessionCost > 0 || manager.totalCost > 0 {
+                Divider()
+                VStack(alignment: .leading, spacing: 2) {
+                    if manager.sessionCost > 0 {
+                        Text("Session: \(manager.sessionAudioSeconds)s — $\(manager.sessionCost, specifier: "%.4f")")
+                    }
+                    Text("Total: \(formatDuration(manager.totalAudioSeconds)) — $\(manager.totalCost, specifier: "%.4f")")
+                }
+                .font(.caption)
+                .foregroundStyle(.secondary)
+                .padding(.horizontal, 8)
+            }
+
             Divider()
 
             Button("Show Transcription") {
@@ -39,4 +52,11 @@ struct MenuBarView: View {
         }
         .padding(.vertical, 4)
     }
+
+    private func formatDuration(_ seconds: Int) -> String {
+        if seconds < 60 { return "\(seconds)s" }
+        let min = seconds / 60
+        let sec = seconds % 60
+        return "\(min)m \(sec)s"
+    }
 }