Add token/sec counter to tui

pkg/tui/chat.go

@@ -66,6 +66,10 @@ type chatModel struct {
 	replyChunkChan  chan string
 	persistence     bool // whether we will save new messages in the conversation
 
+	tokenCount uint
+	startTime  time.Time
+	elapsed    time.Duration
+
 	// ui state
 	focus           focusState
 	wrap            bool     // whether message content is wrapped to viewport width
@@ -282,6 +286,9 @@ func (m chatModel) Update(msg tea.Msg) (chatModel, tea.Cmd) {
 		}
 		m.updateContent()
 		cmds = append(cmds, m.waitForChunk()) // wait for the next chunk
+
+		m.tokenCount++
+		m.elapsed = time.Now().Sub(m.startTime)
 	case msgAssistantReply:
 		// the last reply that was being worked on is finished
 		reply := models.Message(msg)
@@ -693,10 +700,16 @@ func (m *chatModel) footerView() string {
 		saving,
 		segmentStyle.Render(status),
 	}
-	rightSegments := []string{
+	rightSegments := []string{}
-		segmentStyle.Render(fmt.Sprintf("Model: %s", *m.ctx.Config.Defaults.Model)),
+
+	if m.elapsed > 0 && m.tokenCount > 0 {
+		throughput := fmt.Sprintf("%.0f t/sec", float64(m.tokenCount)/m.elapsed.Seconds())
+		rightSegments = append(rightSegments, segmentStyle.Render(throughput))
 	}
 
+	model := fmt.Sprintf("Model: %s", *m.ctx.Config.Defaults.Model)
+	rightSegments = append(rightSegments, segmentStyle.Render(model))
+
 	left := strings.Join(leftSegments, segmentSeparator)
 	right := strings.Join(rightSegments, segmentSeparator)
 
@@ -857,6 +870,10 @@ func (m *chatModel) promptLLM() tea.Cmd {
 	m.waitingForReply = true
 	m.status = "Press ctrl+c to cancel"
 
+	m.tokenCount = 0
+	m.startTime = time.Now()
+	m.elapsed = 0
+
 	return func() tea.Msg {
 		completionProvider, err := m.ctx.GetCompletionProvider(*m.ctx.Config.Defaults.Model)
 		if err != nil {