ai 7.0.34 → 7.0.36

package/docs/03-ai-sdk-core/25-settings.mdx

@@ -142,7 +142,8 @@ You can specify the timeout either as a number (milliseconds) or as an object wi
 
 - `totalMs`: The total timeout for the entire call including all steps.
 - `stepMs`: The timeout for each individual step (LLM call). This is useful for multi-step generations where you want to limit the time spent on each step independently.
-- `chunkMs`: The timeout between stream chunks (streaming only). The call will abort if no new chunk is received within this duration. This is useful for detecting stalled streams.
+- `firstChunkMs`: The timeout until the first content-bearing output of each step (streaming only). Text deltas, reasoning deltas, tool-input deltas, generated files, and tool calls satisfy the timeout. Response metadata, stream starts, empty deltas, raw chunks, and transport activity do not satisfy or reset it.
+- `chunkMs`: The timeout between content-bearing output chunks after output has started (streaming only). Non-content chunks do not reset it. This is useful for detecting streams that stall after generation begins.
 - `toolMs`: The default timeout for all tool executions. If a tool takes longer, it aborts and returns a tool-error so the model can respond or retry.
 - `tools`: Per-tool timeout overrides using `{toolName}Ms` keys (e.g. `weatherMs`, `slowApiMs`). Takes precedence over `toolMs`. Tool names are type-checked for autocomplete.
 
@@ -195,7 +196,19 @@ const result = await generateText({
 const result = streamText({
   model: __MODEL__,
   prompt: 'Invent a new holiday and describe its traditions.',
-  timeout: { chunkMs: 5000 }, // abort if no chunk received for 5 seconds
+  timeout: { chunkMs: 5000 }, // abort if content stalls for 5 seconds
+});
+```
+
+#### Example: First-content timeout for streaming (streamText only)
+
+```ts
+const result = streamText({
+  model: __MODEL__,
+  prompt: 'Invent a new holiday and describe its traditions.',
+  timeout: {
+    firstChunkMs: 10000, // 10 seconds for first content in each step
+  },
 });
 ```
 

package/docs/03-ai-sdk-core/65-devtools.mdx

@@ -79,6 +79,12 @@ npx @ai-sdk/devtools@latest
 
 Open [http://localhost:4983](http://localhost:4983) to view your AI SDK interactions.
 
+### Choose a viewer theme
+
+The viewer uses the dark theme by default. Use the theme button in the viewer
+header to switch between dark and light themes. Your selection is stored in
+your browser and restored the next time you open the viewer at the same origin.
+
 ### Monorepo usage
 
 If you are using a monorepo setup (e.g. Turborepo, Nx), start DevTools from the same workspace where your AI SDK code runs.

package/docs/03-ai-sdk-harnesses/02-harness-agent.mdx

@@ -192,6 +192,46 @@ const result = await agent.continueStream({ session });
 Use `continueStream()` for incremental output, or `continueGenerate()` to drain
 the continued turn and return a `GenerateTextResult`.
 
+## Stop After a Harness Step
+
+Use `stopWhen` to opt into semantic step boundaries. Predicates run after real
+harness tool steps that can continue into another model step, and receive the
+completed steps from the current invocation. When a predicate matches, the
+returned result finishes while the underlying turn remains unfinished. A
+terminal text-only step instead consumes the turn's `finish` event and finishes
+naturally.
+
+```ts
+import { isStepCount } from 'ai';
+
+const steppedAgent = new HarnessAgent({
+  harness: claudeCode,
+  sandbox: createVercelSandbox({
+    runtime: 'node24',
+    ports: [4000],
+  }),
+  stopWhen: isStepCount(1),
+});
+
+const session = await steppedAgent.createSession();
+const result = await steppedAgent.generate({
+  session,
+  prompt: 'Create a short TODO.md for this repository.',
+});
+
+if (session.hasUnfinishedTurn()) {
+  const continueFrom = await session.suspendTurn();
+  await persistContinuationState({ chatId, continuationState: continueFrom });
+}
+```
+
+`stopWhen` has no default. When omitted, `HarnessAgent` continues running until
+the turn naturally finishes or pauses for host input, preserving the behavior
+of agents without step control. Pass one predicate or an array; matching any
+predicate finishes the current result slice. Resume a stopped turn with
+`createSession({ continueFrom })` and `continueStream()` or
+`continueGenerate()`.
+
 ## Prepare the Sandbox
 
 Use `sandboxConfig` to prepare the sandbox before the harness starts.
@@ -325,6 +365,8 @@ console.log(preparation.identity);
 - `sandbox`: a `HarnessV1SandboxProvider`.
 - `id`: optional stable agent identifier.
 - `instructions`: instructions applied once to a fresh session.
+- `stopWhen`: condition(s) for finishing a result slice after a completed
+  harness tool step that can continue into another model step.
 - `tools`: AI SDK tools executed by the host when the harness calls them.
 - `activeTools`: allowlist of built-in and host-executed tools the harness can
   call.

package/docs/04-ai-sdk-ui/03-chatbot-tool-usage.mdx

@@ -128,6 +128,11 @@ There are three things worth mentioning:
    It asks the user for confirmation and displays the result once the user confirms or denies the execution.
    The result is added to the chat using `addToolOutput` with the `tool` parameter for type safety.
 
+Typed tool parts also include the `approval-requested`, `approval-responded`,
+and `output-denied` states. Include these states when handling `part.state`
+exhaustively, even when a tool does not require approval. See
+[Tool execution approval](#tool-execution-approval) for a complete approval UI.
+
 ```tsx filename='app/page.tsx' highlight="6,11,16,19-23,25-26,28-33,51,66-70,77-81"
 'use client';
 
@@ -217,6 +222,10 @@ export default function Chat() {
                         </div>
                       </div>
                     );
+                  case 'approval-requested':
+                    return <div key={callId}>Approval requested.</div>;
+                  case 'approval-responded':
+                    return <div key={callId}>Approval response received.</div>;
                   case 'output-available':
                     return (
                       <div key={callId}>
@@ -225,6 +234,8 @@ export default function Chat() {
                     );
                   case 'output-error':
                     return <div key={callId}>Error: {part.errorText}</div>;
+                  case 'output-denied':
+                    return <div key={callId}>Tool call denied.</div>;
                 }
                 break;
               }
@@ -239,6 +250,10 @@ export default function Chat() {
                     );
                   case 'input-available':
                     return <div key={callId}>Getting location...</div>;
+                  case 'approval-requested':
+                    return <div key={callId}>Approval requested.</div>;
+                  case 'approval-responded':
+                    return <div key={callId}>Approval response received.</div>;
                   case 'output-available':
                     return <div key={callId}>Location: {part.output}</div>;
                   case 'output-error':
@@ -247,6 +262,8 @@ export default function Chat() {
                         Error getting location: {part.errorText}
                       </div>
                     );
+                  case 'output-denied':
+                    return <div key={callId}>Location request denied.</div>;
                 }
                 break;
               }
@@ -266,6 +283,10 @@ export default function Chat() {
                         Getting weather information for {part.input.city}...
                       </div>
                     );
+                  case 'approval-requested':
+                    return <div key={callId}>Approval requested.</div>;
+                  case 'approval-responded':
+                    return <div key={callId}>Approval response received.</div>;
                   case 'output-available':
                     return (
                       <div key={callId}>
@@ -279,6 +300,8 @@ export default function Chat() {
                         {part.errorText}
                       </div>
                     );
+                  case 'output-denied':
+                    return <div key={callId}>Weather request denied.</div>;
                 }
                 break;
               }
@@ -631,10 +654,16 @@ export default function Chat() {
                     return <pre>{JSON.stringify(part.input, null, 2)}</pre>;
                   case 'input-available':
                     return <pre>{JSON.stringify(part.input, null, 2)}</pre>;
+                  case 'approval-requested':
+                    return <div>Approval requested.</div>;
+                  case 'approval-responded':
+                    return <div>Approval response received.</div>;
                   case 'output-available':
                     return <pre>{JSON.stringify(part.output, null, 2)}</pre>;
                   case 'output-error':
                     return <div>Error: {part.errorText}</div>;
+                  case 'output-denied':
+                    return <div>Tool call denied.</div>;
                 }
             }
           })}

package/docs/04-ai-sdk-ui/21-error-handling.mdx

@@ -111,11 +111,13 @@ export default function Chat() {
 }
 ```
 
-#### Alternative: replace last message
+#### Alternative: replace the failed message
 
-Alternatively you can write a custom submit handler that replaces the last message when an error is present.
+Alternatively, you can write a custom submit handler that replaces the failed
+user message with new input. If the assistant response started streaming before
+the error, remove both the partial assistant response and its user message.
 
-```tsx file="app/page.tsx" highlight="13-15,17-18,35"
+```tsx file="app/page.tsx" highlight="13-19,21-22,39"
 'use client';
 
 import { useChat } from '@ai-sdk/react';
@@ -129,7 +131,11 @@ export default function Chat() {
     event.preventDefault();
 
     if (error != null) {
-      setMessages(messages.slice(0, -1)); // remove last message
+      setMessages(messages =>
+        messages.at(-1)?.role === 'assistant'
+          ? messages.slice(0, -2)
+          : messages.slice(0, -1),
+      );
     }
 
     sendMessage({ text: input });

package/docs/07-reference/01-ai-sdk-core/02-stream-text.mdx

@@ -477,10 +477,10 @@ To see `streamText` in action, check out [these examples](#examples).
     },
     {
       name: 'timeout',
-      type: 'number | { totalMs?: number; stepMs?: number; chunkMs?: number; toolMs?: number; tools?: { [toolName]Ms?: number } }',
+      type: 'number | { totalMs?: number; stepMs?: number; firstChunkMs?: number; chunkMs?: number; toolMs?: number; tools?: { [toolName]Ms?: number } }',
       isOptional: true,
       description:
-        'Timeout in milliseconds. Can be specified as a number or as an object with totalMs, stepMs, chunkMs, toolMs, and/or tools properties. totalMs sets the total timeout for the entire call. stepMs sets the timeout for each individual step (LLM call). chunkMs sets the timeout between stream chunks - the call will abort if no new chunk is received within this duration. toolMs sets the default timeout for all tool executions. tools sets per-tool timeout overrides using the pattern {toolName}Ms (e.g. weatherMs, slowApiMs) that take precedence over toolMs - tool names are type-checked for autocomplete. If a tool takes longer than its timeout, it aborts and returns a tool-error so the model can respond or retry. Can be used alongside abortSignal.',
+        'Timeout in milliseconds. Can be specified as a number or as an object with totalMs, stepMs, firstChunkMs, chunkMs, toolMs, and/or tools properties. totalMs sets the total timeout for the entire call. stepMs sets the timeout for each individual step (LLM call). firstChunkMs sets the timeout until the first content-bearing output in each step; metadata, stream starts, empty deltas, raw chunks, and transport activity do not satisfy it. chunkMs sets the timeout between content-bearing output chunks after output has started; non-content chunks do not reset it. toolMs sets the default timeout for all tool executions. tools sets per-tool timeout overrides using the pattern {toolName}Ms (e.g. weatherMs, slowApiMs) that take precedence over toolMs - tool names are type-checked for autocomplete. If a tool takes longer than its timeout, it aborts and returns a tool-error so the model can respond or retry. Can be used alongside abortSignal.',
     },
     {
       name: 'headers',
@@ -4020,7 +4020,7 @@ To see `streamText` in action, check out [these examples](#examples).
     },
     {
       name: 'pipeUIMessageStreamToResponse',
-      type: '(response: ServerResponse, options?: ResponseInit & UIMessageStreamOptions) => void',
+      type: '(response: ServerResponse, options?: ResponseInit & UIMessageStreamOptions) => Promise<void>',
       description:
         'Writes UI message stream output to a Node.js response-like object.',
       properties: [
@@ -4051,7 +4051,7 @@ To see `streamText` in action, check out [these examples](#examples).
     },
     {
       name: 'pipeTextStreamToResponse',
-      type: '(response: ServerResponse, init?: ResponseInit) => void',
+      type: '(response: ServerResponse, init?: ResponseInit) => Promise<void>',
       description:
         'Writes text delta output to a Node.js response-like object. It sets a `Content-Type` header to `text/plain; charset=utf-8` and writes each text delta as a separate chunk.',
       properties: [

package/docs/07-reference/01-ai-sdk-core/16-tool-loop-agent.mdx

@@ -246,8 +246,9 @@ To see `ToolLoopAgent` in action, check out [these examples](#examples).
             },
             {
               name: 'timeout',
-              type: 'number | { totalMs?: number; stepMs?: number; chunkMs?: number } | undefined',
-              description: 'Timeout configuration for the generation.',
+              type: 'number | { totalMs?: number; stepMs?: number; firstChunkMs?: number; chunkMs?: number } | undefined',
+              description:
+                'Timeout configuration for the generation. firstChunkMs and chunkMs are only enforced by stream().',
             },
             {
               name: 'headers',
@@ -360,8 +361,9 @@ To see `ToolLoopAgent` in action, check out [these examples](#examples).
             },
             {
               name: 'timeout',
-              type: 'number | { totalMs?: number; stepMs?: number; chunkMs?: number } | undefined',
-              description: 'Timeout configuration for the generation.',
+              type: 'number | { totalMs?: number; stepMs?: number; firstChunkMs?: number; chunkMs?: number } | undefined',
+              description:
+                'Timeout configuration for the generation. firstChunkMs and chunkMs are only enforced by stream().',
             },
             {
               name: 'headers',
@@ -712,10 +714,10 @@ const result = await agent.generate({
     },
     {
       name: 'timeout',
-      type: 'number | { totalMs?: number; stepMs?: number; chunkMs?: number }',
+      type: 'number | { totalMs?: number; stepMs?: number; firstChunkMs?: number; chunkMs?: number }',
       isOptional: true,
       description:
-        'Timeout in milliseconds. Can be specified as a number or as an object with totalMs, stepMs, and/or chunkMs properties. The call will be aborted if it takes longer than the specified timeout. Can be used alongside abortSignal.',
+        'Timeout in milliseconds. Can be specified as a number or as an object with totalMs, stepMs, firstChunkMs, and/or chunkMs properties. firstChunkMs and chunkMs have no effect on generate(); they are streaming-only and are enforced by stream(). Can be used alongside abortSignal.',
     },
     {
       name: 'experimental_sandbox',
@@ -828,10 +830,10 @@ for await (const chunk of stream.textStream) {
     },
     {
       name: 'timeout',
-      type: 'number | { totalMs?: number; stepMs?: number; chunkMs?: number }',
+      type: 'number | { totalMs?: number; stepMs?: number; firstChunkMs?: number; chunkMs?: number }',
       isOptional: true,
       description:
-        'Timeout in milliseconds. Can be specified as a number or as an object with totalMs, stepMs, and/or chunkMs properties. The call will be aborted if it takes longer than the specified timeout. Can be used alongside abortSignal.',
+        'Timeout in milliseconds. Can be specified as a number or as an object with totalMs, stepMs, firstChunkMs, and/or chunkMs properties. firstChunkMs limits the wait for the first content-bearing output in each model-call step. chunkMs limits gaps between later content-bearing output chunks. Can be used alongside abortSignal.',
     },
     {
       name: 'experimental_sandbox',

package/docs/07-reference/02-ai-sdk-ui/42-pipe-ui-message-stream-to-response.mdx

@@ -17,7 +17,7 @@ The `pipeUIMessageStreamToResponse` function pipes streaming data to a Node.js S
 ## Example
 
 ```tsx
-pipeUIMessageStreamToResponse({
+await pipeUIMessageStreamToResponse({
   response: serverResponse,
   status: 200,
   statusText: 'OK',
@@ -75,3 +75,8 @@ pipeUIMessageStreamToResponse({
     },
   ]}
 />
+
+### Returns
+
+A `Promise<void>` that resolves when the stream has been written to the response
+and rejects when reading or writing the stream fails.

package/docs/08-migration-guides/24-migration-guide-6-0.mdx

@@ -525,6 +525,34 @@ The `unknown` finish reason has been removed. It is now returned as `other`.
 
 ## AI SDK UI
 
+### Tool UI Part Approval States
+
+AI SDK 6 adds `approval-requested`, `approval-responded`, and `output-denied`
+to the tool UI part `state` union. Update exhaustive `switch` statements and
+other state handling to cover the three approval states.
+
+```tsx filename="AI SDK 6"
+switch (part.state) {
+  case 'input-streaming':
+    return 'Loading input';
+  case 'input-available':
+    return 'Input ready';
+  case 'approval-requested':
+    return 'Approval requested';
+  case 'approval-responded':
+    return 'Approval response received';
+  case 'output-available':
+    return 'Output ready';
+  case 'output-error':
+    return part.errorText;
+  case 'output-denied':
+    return 'Tool call denied';
+}
+```
+
+See [Tool execution approval](/docs/ai-sdk-ui/chatbot-tool-usage#tool-execution-approval)
+for handling approval requests and responses in a chat UI.
+
 ### Tool UI Part Helper Functions Rename
 
 The tool UI part helper functions have been renamed to better reflect their purpose and to accommodate both static and dynamic tool parts ([PR #XXXX](https://github.com/vercel/ai/pull/XXXX)).

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai",
-  "version": "7.0.34",
+  "version": "7.0.36",
   "type": "module",
   "description": "AI SDK by Vercel - build apps like ChatGPT, Claude, Gemini, and more with a single interface for any model using the Vercel AI Gateway or go direct to OpenAI, Anthropic, Google, or any other model provider.",
   "license": "Apache-2.0",
@@ -42,7 +42,7 @@
     }
   },
   "dependencies": {
-    "@ai-sdk/gateway": "4.0.26",
+    "@ai-sdk/gateway": "4.0.27",
     "@ai-sdk/provider": "4.0.3",
     "@ai-sdk/provider-utils": "5.0.12"
   },

package/src/agent/agent.ts

@@ -70,7 +70,10 @@ export type AgentCallParameters<
     abortSignal?: AbortSignal;
 
     /**
-     * Timeout in milliseconds. Can be specified as a number or as an object with `totalMs`.
+     * Timeout in milliseconds. Can be specified as a number or as an object
+     * with total, per-step, first-content, inter-content, and tool timeouts.
+     * First-content and inter-content timeouts are only enforced by streaming
+     * calls.
      */
     timeout?: TimeoutConfiguration<TOOLS>;
 

package/src/agent/pipe-agent-ui-stream-to-response.ts

@@ -66,7 +66,7 @@ export async function pipeAgentUIStreamToResponse<
   UIMessageStreamOptions<
     UIMessage<MESSAGE_METADATA, never, InferUITools<TOOLS>>
   >): Promise<void> {
-  pipeUIMessageStreamToResponse({
+  return pipeUIMessageStreamToResponse({
     response,
     headers,
     status,

package/src/generate-object/stream-object-result.ts

@@ -85,7 +85,10 @@ export interface StreamObjectResult<PARTIAL, RESULT, ELEMENT_STREAM> {
    * @param response A Node.js response-like object (ServerResponse).
    * @param init Optional headers, status code, and status text.
    */
-  pipeTextStreamToResponse(response: ServerResponse, init?: ResponseInit): void;
+  pipeTextStreamToResponse(
+    response: ServerResponse,
+    init?: ResponseInit,
+  ): Promise<void>;
 
   /**
    * Creates a simple text stream response.

package/src/generate-object/stream-object.ts

@@ -964,7 +964,7 @@ class DefaultStreamObjectResult<
   }
 
   pipeTextStreamToResponse(response: ServerResponse, init?: ResponseInit) {
-    pipeTextStreamToResponse({
+    return pipeTextStreamToResponse({
       response,
       stream: this.textStream,
       ...init,

package/src/generate-text/generate-text-events.ts

@@ -56,7 +56,8 @@ export type GenerateTextStartEvent<
 
   /**
    * Timeout configuration for the generation.
-   * Can be a number (milliseconds) or an object with totalMs, stepMs, chunkMs, toolMs, and per-tool overrides via tools.
+   * Can be a number (milliseconds) or an object with totalMs, stepMs,
+   * firstChunkMs (streaming only), chunkMs, toolMs, and per-tool overrides via tools.
    */
   readonly timeout: TimeoutConfiguration<TOOLS> | undefined;
 

package/src/generate-text/stream-text-result.ts

@@ -381,7 +381,7 @@ export interface StreamTextResult<
   pipeUIMessageStreamToResponse<UI_MESSAGE extends UIMessage>(
     response: ServerResponse,
     options?: UIMessageStreamResponseInit & UIMessageStreamOptions<UI_MESSAGE>,
-  ): void;
+  ): Promise<void>;
 
   /**
    * Writes text delta output to a Node.js response-like object.
@@ -395,7 +395,10 @@ export interface StreamTextResult<
    *   `pipeTextStreamToResponse` helpers from `'ai'` with `result.stream`
    *   instead. This method will be removed in the next major release.
    */
-  pipeTextStreamToResponse(response: ServerResponse, init?: ResponseInit): void;
+  pipeTextStreamToResponse(
+    response: ServerResponse,
+    init?: ResponseInit,
+  ): Promise<void>;
 
   /**
    * Converts the result to a streamed response object with a stream data part stream.