diff --git a/web/src/components/AssistantChat/messages/MessageActions.test.tsx b/web/src/components/AssistantChat/messages/MessageActions.test.tsx
index e47f1d13..478f6899 100644
--- a/web/src/components/AssistantChat/messages/MessageActions.test.tsx
+++ b/web/src/components/AssistantChat/messages/MessageActions.test.tsx
@@ -58,7 +58,7 @@ describe('MessageActions', () => {
         expect(screen.getByRole('button', { name: 'Message details' })).toBeTruthy()
         expect(screen.getByText('Duration: 1.3s')).toBeTruthy()
         expect(screen.getByText('Model: gpt-5.2-codex')).toBeTruthy()
-        expect(screen.getByText('Usage: 125 billable tokens (100 in / 25 out)')).toBeTruthy()
+        expect(screen.getByText('Tokens: 125 total (100 in / 25 out)')).toBeTruthy()
         expect(screen.queryByText(/^Invoke:/)).toBeNull()
     })
 
diff --git a/web/src/components/AssistantChat/messages/MessageMetadata.test.ts b/web/src/components/AssistantChat/messages/MessageMetadata.test.ts
index f44274b5..36a1b62f 100644
--- a/web/src/components/AssistantChat/messages/MessageMetadata.test.ts
+++ b/web/src/components/AssistantChat/messages/MessageMetadata.test.ts
@@ -41,12 +41,12 @@ describe('buildMessageMetadataLabels', () => {
         expect(buildMessageMetadataLabels({})).toEqual([])
     })
 
-    it('labels token totals as billable to clarify that cache I/O is intentionally excluded', () => {
+    it('renders the token total and input/output breakdown without billing claims', () => {
         const parts = buildMessageMetadataLabels({
             usage: { input_tokens: 100, output_tokens: 200 }
         })
-        expect(parts.some(p => /\bbillable tokens\b/.test(p))).toBe(true)
-        expect(parts.some(p => p.includes('300 billable tokens (100 in / 200 out)'))).toBe(true)
+        expect(parts).toContain('Tokens: 300 total (100 in / 200 out)')
+        expect(parts.some(p => /\bbillable\b/.test(p))).toBe(false)
     })
 
     it('does not drop a Duration line when durationMs is exactly 0', () => {
@@ -83,7 +83,7 @@ describe('buildMessageMetadataLabels', () => {
         expect(parts).toEqual([
             'Duration: 1.2s',
             'Model: claude-sonnet-4-6',
-            'Usage: 22 billable tokens (3 in / 19 out)'
+            'Tokens: 22 total (3 in / 19 out)'
         ])
     })
 
@@ -95,8 +95,7 @@ describe('buildMessageMetadataLabels', () => {
         })
         expect(parts).toContain('Models: claude-sonnet-4-6, claude-haiku-4-5-20251001')
         expect(parts.some(p => p.startsWith('Model:'))).toBe(false)
-        expect(parts).toContain('Total: 300 billable tokens (100 in / 200 out)')
-        expect(parts.some(p => p.startsWith('Usage:'))).toBe(false)
+        expect(parts).toContain('Tokens: 300 total (100 in / 200 out)')
         expect(parts).toContain('3 turns')
     })
 
diff --git a/web/src/components/AssistantChat/messages/MessageMetadata.tsx b/web/src/components/AssistantChat/messages/MessageMetadata.tsx
index bb73aca3..e2b6b375 100644
--- a/web/src/components/AssistantChat/messages/MessageMetadata.tsx
+++ b/web/src/components/AssistantChat/messages/MessageMetadata.tsx
@@ -37,14 +37,9 @@ export function buildMessageMetadataLabels({ durationMs, usage, model, turnCount
     }
 
     if (usage) {
-        // "Billable" because cache_read tokens are not part of the input
-        // figure used for billing; surfacing only input + output here is
-        // intentional. If we ever surface cache I/O, keep this label and
-        // add a separate `Cache:` line.
         const total = usage.input_tokens + usage.output_tokens
         const formatToken = (n: number) => n.toLocaleString()
-        const usageLabel = isAggregated ? 'Total' : 'Usage'
-        parts.push(`${usageLabel}: ${formatToken(total)} billable tokens (${formatToken(usage.input_tokens)} in / ${formatToken(usage.output_tokens)} out)`)
+        parts.push(`Tokens: ${formatToken(total)} total (${formatToken(usage.input_tokens)} in / ${formatToken(usage.output_tokens)} out)`)
     }
 
     if (isAggregated) {