AlemTuzlak · Oct 5, 2026 · Oct 5, 2026
diff --git a/.changeset/chat-hydrate-strict-mode.md b/.changeset/chat-hydrate-strict-mode.md
diff --git a/.changeset/clean-stream-readers.md b/.changeset/clean-stream-readers.md
diff --git a/.changeset/code-mode-stubs-enum-description.md b/.changeset/code-mode-stubs-enum-description.md
diff --git a/.changeset/durable-batch-live-flush.md b/.changeset/durable-batch-live-flush.md
diff --git a/.changeset/fix-malformed-chat-completions-tool-arguments.md b/.changeset/fix-malformed-chat-completions-tool-arguments.md
diff --git a/.changeset/openrouter-optional-tool-fields.md b/.changeset/openrouter-optional-tool-fields.md
diff --git a/.changeset/sync-models.md b/.changeset/sync-models.md
diff --git a/.changeset/tidy-boxd-cancellation.md b/.changeset/tidy-boxd-cancellation.md
diff --git a/examples/ag-ui/CHANGELOG.md b/examples/ag-ui/CHANGELOG.md
@@ -1,5 +1,13 @@
 # ag-ui
 
+## 0.0.32
+
+### Patch Changes
+
+- Updated dependencies [[`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929), [`e436250`](https://github.com/TanStack/ai/commit/e4362509fd7becad7175ce895178cd3b107fcd4b)]:
+  - @tanstack/ai-client@0.36.2
+  - @tanstack/ai-react@0.29.4
+
 ## 0.0.31
 
 ### Patch Changes

diff --git a/examples/ag-ui/package.json b/examples/ag-ui/package.json
@@ -2,7 +2,7 @@
   "name": "ag-ui",
   "private": true,
   "type": "module",
-  "version": "0.0.31",
+  "version": "0.0.32",
   "scripts": {
     "detect-servers": "node scripts/detect-servers.mjs",
     "predev": "node scripts/detect-servers.mjs",

diff --git a/packages/ai-client/CHANGELOG.md b/packages/ai-client/CHANGELOG.md
@@ -1,5 +1,20 @@
 # @tanstack/ai-client
 
+## 0.36.2
+
+### Patch Changes
+
+- [#1620](https://github.com/TanStack/ai/pull/1620) [`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929) - Send one hydrate `GET` when a chat with `persistence: true` mounts in React Strict Mode. Strict Mode attaches, detaches and attaches the client again in dev, and each attach sent its own `GET`. A re-attach now reuses the `GET` that is still in flight.
+
+- [#1598](https://github.com/TanStack/ai/pull/1598) [`e436250`](https://github.com/TanStack/ai/commit/e4362509fd7becad7175ce895178cd3b107fcd4b) - Cancel fetch response bodies when SSE or NDJSON parsing fails, the consumer exits early, or SSE reaches a `[DONE]` marker. This closes unfinished connections and preserves the original error if cancellation fails.
+
+  Also cancel streaming generation responses when the client stops reading, including after a `RUN_ERROR` event.
+
+  Start cancellation without waiting for custom hooks, so pending hooks cannot delay errors, iterator completion, or reader lock release.
+
+- Updated dependencies [[`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929)]:
+  - @tanstack/ai@0.64.1
+
 ## 0.36.1
 
 ### Patch Changes

diff --git a/packages/ai-client/package.json b/packages/ai-client/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai-client",
-  "version": "0.36.1",
+  "version": "0.36.2",
   "description": "Framework-agnostic headless client for TanStack AI chat, realtime sessions, streaming transports, and media generations.",
   "author": "Tanner Linsley",
   "license": "MIT",

diff --git a/packages/ai-code-mode/CHANGELOG.md b/packages/ai-code-mode/CHANGELOG.md
@@ -1,5 +1,17 @@
 # @tanstack/ai-code-mode
 
+## 0.4.20
+
+### Patch Changes
+
+- [#1600](https://github.com/TanStack/ai/pull/1600) [`b8bab43`](https://github.com/TanStack/ai/commit/b8bab43b98ac91e201dc1dae8a3811b348d34504) - Keep `enum` and `const` values and property descriptions in the generated
+  Code Mode type stubs. A typed schema such as `{ type: 'string', enum: ['a', 'b'] }`
+  now renders as `'a' | 'b'` instead of `string`, and property `description`s are
+  emitted as JSDoc, as the `includeDescriptions` option of `generateTypeStubs`
+  already documented.
+- Updated dependencies [[`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929)]:
+  - @tanstack/ai@0.64.1
+
 ## 0.4.19
 
 ### Patch Changes

diff --git a/packages/ai-code-mode/package.json b/packages/ai-code-mode/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai-code-mode",
-  "version": "0.4.19",
+  "version": "0.4.20",
   "description": "Secure TypeScript Code Mode for TanStack AI agents to execute sandboxed tool orchestration programs.",
   "author": "Tanner Linsley",
   "license": "MIT",

diff --git a/packages/ai-octane/CHANGELOG.md b/packages/ai-octane/CHANGELOG.md
@@ -1,5 +1,13 @@
 # @tanstack/ai-octane
 
+## 0.6.5
+
+### Patch Changes
+
+- Updated dependencies [[`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929), [`e436250`](https://github.com/TanStack/ai/commit/e4362509fd7becad7175ce895178cd3b107fcd4b), [`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929)]:
+  - @tanstack/ai-client@0.36.2
+  - @tanstack/ai@0.64.1
+
 ## 0.6.4
 
 ### Patch Changes

diff --git a/packages/ai-octane/package.json b/packages/ai-octane/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai-octane",
-  "version": "0.6.4",
+  "version": "0.6.5",
   "description": "Octane bindings for TanStack AI streaming chat, structured outputs, and media generation.",
   "author": "Dominic Gannaway",
   "license": "MIT",

diff --git a/packages/ai-openrouter/CHANGELOG.md b/packages/ai-openrouter/CHANGELOG.md
@@ -1,5 +1,16 @@
 # @tanstack/ai-openrouter
 
+## 0.20.3
+
+### Patch Changes
+
+- [#1545](https://github.com/TanStack/ai/pull/1545) [`aadfed3`](https://github.com/TanStack/ai/commit/aadfed3cff1925f647cbf956f309f0a38b7ccf85) - Keep optional tool fields optional. The Chat Completions adapter now sends function tools with `strict: false`. With `strict` omitted, OpenAI models through OpenRouter were observed to treat the schema as strict and make every optional field required, so before this fix the model could not leave one out. The Responses adapter now strips the `null` that strict mode puts in an omitted optional field before the tool input is validated, so the tool runs and sees the field as absent. A `.nullable()` field keeps its `null`. Known limit: an optional field inside a union of several object shapes is not stripped and still arrives as `null`.
+
+- [#1576](https://github.com/TanStack/ai/pull/1576) [`0977222`](https://github.com/TanStack/ai/commit/097722249c033792538d6f95f5db76473a6e3035) - Update model metadata from OpenRouter API
+
+- Updated dependencies [[`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929)]:
+  - @tanstack/ai@0.64.1
+
 ## 0.20.2
 
 ### Patch Changes

diff --git a/packages/ai-openrouter/package.json b/packages/ai-openrouter/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai-openrouter",
-  "version": "0.20.2",
+  "version": "0.20.3",
   "description": "TanStack AI adapter for OpenRouter chat, provider tools, structured outputs, and access to hundreds of LLMs.",
   "author": "Tanner Linsley",
   "license": "MIT",

diff --git a/packages/ai-sandbox-boxd/CHANGELOG.md b/packages/ai-sandbox-boxd/CHANGELOG.md
@@ -1,5 +1,14 @@
 # @tanstack/ai-sandbox-boxd
 
+## 0.2.1
+
+### Patch Changes
+
+- [#1607](https://github.com/TanStack/ai/pull/1607) [`7216c7b`](https://github.com/TanStack/ai/commit/7216c7b5813074eafad52028b3d6d7fec1f7d08c) - Delete newly created machines when create or snapshot restore is aborted during readiness or workspace setup.
+
+- Updated dependencies []:
+  - @tanstack/ai-sandbox@0.5.18
+
 ## 0.2.0
 
 ### Minor Changes

diff --git a/packages/ai-sandbox-boxd/package.json b/packages/ai-sandbox-boxd/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai-sandbox-boxd",
-  "version": "0.2.0",
+  "version": "0.2.1",
   "description": "boxd sandbox provider for TanStack AI — run harness adapters inside isolated boxd KVM microVMs through the uniform SandboxHandle.",
   "author": "",
   "license": "MIT",

diff --git a/packages/ai-vercel-gateway/CHANGELOG.md b/packages/ai-vercel-gateway/CHANGELOG.md
@@ -1,5 +1,15 @@
 # @tanstack/ai-vercel-gateway
 
+## 0.3.3
+
+### Patch Changes
+
+- [#1576](https://github.com/TanStack/ai/pull/1576) [`0977222`](https://github.com/TanStack/ai/commit/097722249c033792538d6f95f5db76473a6e3035) - Update model metadata from OpenRouter API
+
+- Updated dependencies [[`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929), [`6d8e648`](https://github.com/TanStack/ai/commit/6d8e6485f92c98a6f2200471e2aff21ab50be013)]:
+  - @tanstack/ai@0.64.1
+  - @tanstack/openai-base@0.12.3
+
 ## 0.3.2
 
 ### Patch Changes

diff --git a/packages/ai-vercel-gateway/package.json b/packages/ai-vercel-gateway/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai-vercel-gateway",
-  "version": "0.3.2",
+  "version": "0.3.3",
   "description": "Vercel AI Gateway adapter for TanStack AI chat, embeddings, and image generation.",
   "author": "Tanner Linsley",
   "license": "MIT",

diff --git a/packages/ai/CHANGELOG.md b/packages/ai/CHANGELOG.md
@@ -1,5 +1,13 @@
 # @tanstack/ai
 
+## 0.64.1
+
+### Patch Changes
+
+- [#1620](https://github.com/TanStack/ai/pull/1620) [`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929) - Stream durable responses live again. With `durability` set, a chunk waited in the append batch until 32 chunks arrived or the run finished, so a short reply showed up all at once at the end. Now the batch also flushes when the model sends no new chunk for 50ms. You do not need `batch: 1` for live text any more. `batch` stays the largest number of chunks in one append.
+
+  Set the wait with the new `batchWaitMs` option on `durability` (`toServerSentEventsResponse`, `toHttpResponse`) and on `toWebSocketStream` / `toWebSocketResponse`. A higher value means fewer writes to the log but slower live text. `0` appends every chunk on its own.
+
 ## 0.64.0
 
 ### Minor Changes

diff --git a/packages/ai/package.json b/packages/ai/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai",
-  "version": "0.64.0",
+  "version": "0.64.1",
   "description": "Type-safe TypeScript AI SDK for streaming chat, tool calling, agents, structured outputs, and multimodal generation.",
   "author": "Tanner Linsley",
   "license": "MIT",

diff --git a/packages/openai-base/CHANGELOG.md b/packages/openai-base/CHANGELOG.md
@@ -1,5 +1,14 @@
 # @tanstack/openai-base
 
+## 0.12.3
+
+### Patch Changes
+
+- [#1602](https://github.com/TanStack/ai/pull/1602) [`6d8e648`](https://github.com/TanStack/ai/commit/6d8e6485f92c98a6f2200471e2aff21ab50be013) - Preserve malformed Chat Completions tool arguments so they return a tool error without executing the tool with an empty object.
+
+- Updated dependencies [[`4c57d04`](https://github.com/TanStack/ai/commit/4c57d04f0f5dd98e176386f3208b68daf9e7d929)]:
+  - @tanstack/ai@0.64.1
+
 ## 0.12.2
 
 ### Patch Changes

diff --git a/packages/openai-base/package.json b/packages/openai-base/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/openai-base",
-  "version": "0.12.2",
+  "version": "0.12.3",
   "description": "Shared OpenAI SDK base adapters for TanStack AI providers using Chat Completions and Responses APIs.",
   "author": "Tanner Linsley",
   "license": "MIT",