fix(usage): apply flat Sonnet 4.6 pricing

src/main/claude-usage/claude-model-pricing.test.ts

@@ -22,19 +22,23 @@ describe('estimateCostUsd cache-write TTL rates', () => {
     expect(estimateCostUsd('claude-opus-5', 0, 0, 0, 1_000, 5_000)).toBeCloseTo(0.01)
   })
 
-  it('applies the long-context tier to 1-hour writes', () => {
-    expect(estimateCostUsd('claude-sonnet-4-6', 0, 0, 0, 400_000, 400_000)).toBeCloseTo(3.6)
+  it('keeps Sonnet 4.6 one-hour writes flat across its full 1M window', () => {
+    expect(estimateCostUsd('claude-sonnet-4-6', 0, 0, 0, 400_000, 400_000)).toBeCloseTo(2.4)
   })
 
-  it('shares one long-context allowance across both TTL buckets', () => {
+  it('applies the legacy long-context tier to Sonnet 4.5 one-hour writes', () => {
+    expect(estimateCostUsd('claude-sonnet-4-5', 0, 0, 0, 400_000, 400_000)).toBeCloseTo(3.6)
+  })
+
+  it('shares one legacy long-context allowance across both TTL buckets', () => {
     // 400k writes split evenly: 200k @ (3.75/7.5) and 200k @ (6/12), each tier
     // getting half of the 200k allowance.
-    expect(estimateCostUsd('claude-sonnet-4-6', 0, 0, 0, 400_000, 200_000)).toBeCloseTo(2.925)
+    expect(estimateCostUsd('claude-sonnet-4-5', 0, 0, 0, 400_000, 200_000)).toBeCloseTo(2.925)
   })
 
-  it('never lowers a long-context estimate as writes shift from 5-minute to 1-hour', () => {
-    const costs = [0, 50_000, 100_000, 200_000, 300_000, 400_000].map((write1h) =>
-      estimateCostUsd('claude-sonnet-4-6', 0, 0, 0, 400_000, write1h)!
+  it('never lowers a legacy long-context estimate as writes shift to 1-hour', () => {
+    const costs = [0, 50_000, 100_000, 200_000, 300_000, 400_000].map(
+      (write1h) => estimateCostUsd('claude-sonnet-4-5', 0, 0, 0, 400_000, write1h)!
     )
     for (let index = 1; index < costs.length; index++) {
       expect(costs[index]).toBeGreaterThan(costs[index - 1])

src/main/claude-usage/claude-model-pricing.ts

@@ -37,13 +37,13 @@ const MODEL_PRICING: Record<string, ClaudeModelPricing> = {
   'claude-opus-4-5': { input: 5, output: 25, cacheRead: 0.5, cacheWrite: 6.25, cacheWrite1h: 10 },
   'claude-opus-4-1': { input: 15, output: 75, cacheRead: 1.5, cacheWrite: 18.75, cacheWrite1h: 30 },
   'claude-opus-4': { input: 15, output: 75, cacheRead: 1.5, cacheWrite: 18.75, cacheWrite1h: 30 },
+  // Claude 4.6 and later keep standard rates across the full 1M context window.
   'claude-sonnet-4-6': {
     input: 3,
     output: 15,
     cacheRead: 0.3,
     cacheWrite: 3.75,
-    cacheWrite1h: 6,
-    ...SONNET_LONG_CONTEXT_PRICING
+    cacheWrite1h: 6
   },
   'claude-sonnet-4-5': {
     input: 3,

src/main/claude-usage/store.test.ts

@@ -546,7 +546,7 @@ describe('ClaudeUsageStore', () => {
     expect(summary.estimatedCostUsd).toBeCloseTo(220.5)
   })
 
-  it('prices Sonnet long-context usage with threshold rates', async () => {
+  it('prices Sonnet 4.6 long-context usage at its flat 1M-window rates', async () => {
     const store = createStoreWithState({
       dailyAggregates: [
         {
@@ -569,7 +569,7 @@ describe('ClaudeUsageStore', () => {
 
     const summary = await store.getSummary('orca', '30d')
 
-    expect(summary.estimatedCostUsd).toBeCloseTo(8.07)
+    expect(summary.estimatedCostUsd).toBeCloseTo(6.615)
   })
 
   it('returns automation usage for a single matching worktree session', async () => {