update changelog and context default (#1045)

tests/e2e/fixtures.ts

@@ -141,7 +141,7 @@ export async function mockHermesApi(page: Page, options: MockHermesApiOptions =
     }
 
     if (pathname === '/api/hermes/sessions/context-length') {
-      await route.fulfill(jsonResponse({ context_length: 200000 }))
+      await route.fulfill(jsonResponse({ context_length: 256000 }))
       return
     }
 

tests/server/model-context.test.ts

@@ -71,7 +71,7 @@ describe('getModelContextLength', () => {
 
     const { getModelContextLength } = await loadModelContext()
 
-    expect(getModelContextLength()).toBe(200_000)
+    expect(getModelContextLength()).toBe(256_000)
   })
 
   it('does not scan other providers when the configured provider exists without that model', async () => {
@@ -79,7 +79,7 @@ describe('getModelContextLength', () => {
     writeModelsCache({
       'openai-codex': {
         models: {
-          'gpt-5.4': { limit: { context: 200_000 } },
+          'gpt-5.4': { limit: { context: 256_000 } },
         },
       },
       openai: {
@@ -91,7 +91,7 @@ describe('getModelContextLength', () => {
 
     const { getModelContextLength } = await loadModelContext()
 
-    expect(getModelContextLength()).toBe(200_000)
+    expect(getModelContextLength()).toBe(256_000)
   })
 
   it('uses the configured provider cache entry when the provider matches', async () => {
@@ -241,7 +241,7 @@ describe('getModelContextLength', () => {
 
     const { getModelContextLength } = await loadModelContext()
 
-    expect(getModelContextLength()).toBe(200_000)
+    expect(getModelContextLength()).toBe(256_000)
   })
 
   it('does not trust custom:name alone when the matched custom provider entry points at an unknown proxy url', async () => {
@@ -256,7 +256,7 @@ describe('getModelContextLength', () => {
 
     const { getModelContextLength } = await loadModelContext()
 
-    expect(getModelContextLength()).toBe(200_000)
+    expect(getModelContextLength()).toBe(256_000)
   })
 
   it('does not fall through to a unique global match after a resolved custom:name provider misses in its scoped cache provider', async () => {
@@ -276,7 +276,7 @@ describe('getModelContextLength', () => {
 
     const { getModelContextLength } = await loadModelContext()
 
-    expect(getModelContextLength()).toBe(200_000)
+    expect(getModelContextLength()).toBe(256_000)
   })
 
   it('allows a unique global model-name fallback for unresolved custom providers', async () => {
@@ -321,7 +321,7 @@ describe('getModelContextLength', () => {
 
     const { getModelContextLength } = await loadModelContext()
 
-    expect(getModelContextLength()).toBe(200_000)
+    expect(getModelContextLength()).toBe(256_000)
   })
 
   it('does not guess across multiple cache providers when a custom provider remains unresolved', async () => {
@@ -341,6 +341,6 @@ describe('getModelContextLength', () => {
 
     const { getModelContextLength } = await loadModelContext()
 
-    expect(getModelContextLength()).toBe(200_000)
+    expect(getModelContextLength()).toBe(256_000)
   })
 })

tests/server/run-chat-compression.test.ts

@@ -87,7 +87,7 @@ describe('run chat compression trigger', () => {
     readConfigYamlForProfileMock.mockReset()
 
     getSessionMock.mockReturnValue({ id: 'session-1', profile: 'default' })
-    getModelContextLengthMock.mockReturnValue(200_000)
+    getModelContextLengthMock.mockReturnValue(256_000)
     calcAndUpdateUsageMock.mockResolvedValue({ inputTokens: 1_000, outputTokens: 0 })
     estimateUsageTokensFromMessagesMock.mockReturnValue({ inputTokens: 0, outputTokens: 0 })
     getCompressionSnapshotMock.mockReturnValue(null)
@@ -142,7 +142,7 @@ describe('run chat compression trigger', () => {
     readConfigYamlForProfileMock.mockResolvedValue({
       compression: { threshold: 0.25, target_ratio: 0.1, protect_last_n: 7, protect_first_n: 2 },
     })
-    calcAndUpdateUsageMock.mockResolvedValue({ inputTokens: 60_000, outputTokens: 0 })
+    calcAndUpdateUsageMock.mockResolvedValue({ inputTokens: 70_000, outputTokens: 0 })
     compressorCompressMock.mockResolvedValue({
       messages: [{ role: 'user', content: 'compressed' }],
       meta: {
@@ -212,7 +212,7 @@ describe('run chat compression trigger', () => {
       emit,
       new Map(),
       {},
-      vi.fn(async () => 120_000),
+      vi.fn(async () => 160_000),
     )
 
     expect(history).toEqual([{ role: 'user', content: 'compressed by local context estimate' }])
@@ -356,7 +356,7 @@ describe('run chat compression trigger', () => {
       emit,
       new Map(),
       {},
-      vi.fn(async () => 120_000),
+      vi.fn(async () => 160_000),
       700,
     )
 
@@ -387,7 +387,7 @@ describe('run chat compression trigger', () => {
       emit,
       new Map(),
       {},
-      vi.fn(async () => 120_000),
+      vi.fn(async () => 160_000),
     )).rejects.toBeInstanceOf(ContextWindowTooSmallError)
 
     expect(emit).not.toHaveBeenCalledWith('usage.updated', expect.anything())
@@ -420,7 +420,7 @@ describe('run chat compression trigger', () => {
       vi.fn(),
       new Map(),
       {},
-      vi.fn(async () => 120_000),
+      vi.fn(async () => 160_000),
     )).rejects.toBeInstanceOf(ContextWindowTooSmallError)
 
     expect(compressorCompressMock).not.toHaveBeenCalled()
@@ -443,7 +443,7 @@ describe('run chat compression trigger', () => {
     readConfigYamlForProfileMock.mockResolvedValue({
       compression: { protect_last_n: 5 },
     })
-    calcAndUpdateUsageMock.mockResolvedValue({ inputTokens: 120_000, outputTokens: 0 })
+    calcAndUpdateUsageMock.mockResolvedValue({ inputTokens: 160_000, outputTokens: 0 })
     compressorCompressMock.mockResolvedValue({
       messages: [{ role: 'user', content: 'compressed' }],
       meta: {
@@ -468,8 +468,8 @@ describe('run chat compression trigger', () => {
 
     expect(compressorConstructorMock).toHaveBeenCalledWith({
       config: {
-        triggerTokens: 100_000,
-        summaryBudget: 40_000,
+        triggerTokens: 128_000,
+        summaryBudget: 51_200,
         headMessageCount: 3,
         tailMessageCount: 5,
       },
@@ -544,7 +544,7 @@ describe('run chat compression trigger', () => {
     readConfigYamlForProfileMock.mockResolvedValue({
       compression: { protect_first_n: 2, protect_last_n: 3 },
     })
-    estimateUsageTokensFromMessagesMock.mockReturnValue({ inputTokens: 120_000, outputTokens: 0 })
+    estimateUsageTokensFromMessagesMock.mockReturnValue({ inputTokens: 160_000, outputTokens: 0 })
     compressorCompressMock.mockResolvedValue({
       messages: [{ role: 'user', content: 'updated stale compressed' }],
       meta: {

tests/server/sessions-routes.test.ts

@@ -16,7 +16,7 @@ const listWorkspaceFoldersMock = vi.fn(async (ctx: any) => { ctx.body = { folder
 const usageBatchMock = vi.fn(async (ctx: any) => { ctx.body = {} })
 const usageSingleMock = vi.fn(async (ctx: any) => { ctx.body = { input_tokens: 0, output_tokens: 0 } })
 const usageStatsMock = vi.fn(async (ctx: any) => { ctx.body = { total_input_tokens: 0, total_output_tokens: 0 } })
-const contextLengthMock = vi.fn(async (ctx: any) => { ctx.body = { context_length: 200000 } })
+const contextLengthMock = vi.fn(async (ctx: any) => { ctx.body = { context_length: 256000 } })
 const batchRemoveMock = vi.fn(async (ctx: any) => { ctx.body = { deleted: 1, failed: 0, errors: [] } })
 const exportSessionMock = vi.fn(async (ctx: any) => { ctx.body = JSON.stringify({ id: ctx.params.id }) })