backport v6 by callycodes · Pull Request #5 · AllTheTables/ai

callycodes · 2026-01-06T13:35:34Z

Background

Summary

Manual Verification

Checklist

Tests have been added / updated (for bug fixes / features)
Documentation has been added / updated (for bug fixes / features)
A patch changeset for relevant packages has been added (for bug fixes / features - run pnpm changeset in the project root)
Formatting issues have been fixed (run pnpm prettier-fix in the project root)
I have reviewed this pull request (self-review)

Future Work

Related Issues

## Background as a follow up to this comment: #12403 (comment) ## Checklist - [ ] na ~~Tests have been added / updated (for bug fixes / features)~~ - [x] Documentation has been added / updated (for bug fixes / features) - [ ] na ~~A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root)~~ - [x] I have reviewed this pull request (self-review)

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## ai@6.0.91 ### Patch Changes - Updated dependencies [a433cd3] - @ai-sdk/gateway@3.0.50 ## @ai-sdk/angular@2.0.92 ### Patch Changes - ai@6.0.91 ## @ai-sdk/gateway@3.0.50 ### Patch Changes - a433cd3: chore(provider/gateway): update gateway model settings files ## @ai-sdk/langchain@2.0.97 ### Patch Changes - ai@6.0.91 ## @ai-sdk/llamaindex@2.0.91 ### Patch Changes - ai@6.0.91 ## @ai-sdk/react@3.0.93 ### Patch Changes - ai@6.0.91 ## @ai-sdk/rsc@2.0.91 ### Patch Changes - ai@6.0.91 ## @ai-sdk/svelte@4.0.91 ### Patch Changes - ai@6.0.91 ## @ai-sdk/vue@3.0.91 ### Patch Changes - ai@6.0.91 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background OpenAI's Responses API supports skills natively attached to the shell tool environment. Skills allow extending the shell tool with custom instructions and capabilities, similar to Anthropic's computer use features. Additionally, the shell tool now supports running in a hosted container on OpenAI's infrastructure. This is recent and an enhancement over the initial implementation which only supported "local" execution (implemented in #10810 for AI SDK). See: * https://developers.openai.com/api/docs/guides/tools-shell * https://developers.openai.com/api/docs/guides/tools-skills ## Summary Adds native OpenAI skills support to the shell tool environment for the Responses API via `providerOptions`. Supports both local shell execution and hosted container shell execution. **Key changes:** - Extended `openai.tools.shell()` with `environment` configuration supporting three modes: - `containerAuto`: Automatic container provisioning with optional skills - `containerReference`: Reference to existing container - `local`: Local execution with filesystem-based skills - Skills can be provided as inline base64-encoded zip files (`skillReference` or `inline` types for container environments) or filesystem paths (for local environments) - Added `shell_call_output` handling in the language model stream parser (relevant for container hosted shell) - Fixture-based tests with real API responses - Examples demonstrating both container-based and local skill usage - Expands docs accordingly The implementation is fully backward compatible; existing shell tool usage without skills continues to work unchanged (important note: `environment.type` must remain optional and default to `local`); verified by running existing tests and examples originally introduced. ## Manual Verification Ran example scripts with both container-based and local skills configurations: ```bash # Container-based skills (inline skill from zip file) pnpm tsx examples/ai-functions/src/generate-text/openai-responses-shell-container-skills.ts pnpm tsx examples/ai-functions/src/stream-text/openai-responses-shell-container-skills.ts # Local skills (filesystem-based) pnpm tsx examples/ai-functions/src/generate-text/openai-responses-shell-local-skills.ts pnpm tsx examples/ai-functions/src/stream-text/openai-responses-shell-local-skills.ts ``` Also run the other `examples/ai-functions/src/generate-text/openai-responses-shell-*` and `examples/ai-functions/src/stream-text/openai-responses-shell-*` examples to verify back compat. ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work - Abstract OpenAI Skills and Anthropic computer use into unified specification (AI SDK v7) ## Related Issues Fixes #12487 Builds on top of #10810 --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## @ai-sdk/azure@3.0.31 ### Patch Changes - Updated dependencies [ff12133] - @ai-sdk/openai@3.0.30 ## @ai-sdk/openai@3.0.30 ### Patch Changes - ff12133: feat(provider/openai): support native skills and hosted shell Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

see #12673. Agents shouldn't apply these labels, they are reserved for humans to control releases and backports. I really don't want to accidentally release a breaking change lol

## Background Hindsight is state of the art of Memory for AI Agents and the integration for AI SDK has been used for weeks now by users. ## Summary Added Hindsight to community-provided memory ## Manual Verification  ## Checklist  - [ ] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [ ] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [ ] I have reviewed this pull request (self-review) ## Future Work  ## Related Issues  --------- Co-authored-by: nicoalbanese <gcalbanese96@gmail.com> Co-authored-by: Nico Albanese <49612682+nicoalbanese@users.noreply.github.com>

## Background Google released `gemini-3.1-pro-preview`, which needs to be added to the AI SDK provider packages and documentation. See: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/ ## Summary Adds `gemini-3.1-pro-preview` as a supported model ID to `@ai-sdk/google`, `@ai-sdk/google-vertex`, and `@ai-sdk/gateway`, and updates documentation and examples to reference the new model where it makes sense. Updates `thinkingLevel` docs to note that Gemini 3.1 Pro supports 'low', 'medium', and 'high' levels. ## Manual Verification N/A ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work N/A ## Related Issues N/A

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## ai@6.0.92 ### Patch Changes - Updated dependencies [765b013] - @ai-sdk/gateway@3.0.51 ## @ai-sdk/angular@2.0.93 ### Patch Changes - ai@6.0.92 ## @ai-sdk/gateway@3.0.51 ### Patch Changes - 765b013: feat(provider/google): add support for `gemini-3.1-pro-preview` ## @ai-sdk/google@3.0.30 ### Patch Changes - 765b013: feat(provider/google): add support for `gemini-3.1-pro-preview` ## @ai-sdk/google-vertex@4.0.60 ### Patch Changes - 765b013: feat(provider/google): add support for `gemini-3.1-pro-preview` - Updated dependencies [765b013] - @ai-sdk/google@3.0.30 ## @ai-sdk/langchain@2.0.98 ### Patch Changes - ai@6.0.92 ## @ai-sdk/llamaindex@2.0.92 ### Patch Changes - ai@6.0.92 ## @ai-sdk/react@3.0.94 ### Patch Changes - ai@6.0.92 ## @ai-sdk/rsc@2.0.92 ### Patch Changes - ai@6.0.92 ## @ai-sdk/svelte@4.0.92 ### Patch Changes - ai@6.0.92 ## @ai-sdk/vue@3.0.92 ### Patch Changes - ai@6.0.92 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Summary - Replace path-based `/dist/` check with an explicit `AI_SDK_DEVTOOLS_DEV` env flag for dev mode detection - Prevents false positives in monorepos (NX, Turborepo) where `/dist/` paths are common in the workspace - The `dev:api` script now sets `AI_SDK_DEVTOOLS_DEV=true` automatically; production mode is the default when the flag is unset 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## @ai-sdk/devtools@0.0.15 ### Patch Changes - 4c3a932: Use explicit AI_SDK_DEVTOOLS_DEV env flag for dev mode detection Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background The Gemini 1.5 model family and `gemini-2.0-flash-exp` (which was Google's first multimodal image generation model) are no longer around. ## Summary Updates all examples that use one of them to use an equivalent available model. ## Related PRs #12494

## Background our current telemetry setup was heavily integrated with OTel. this is the first step in decoupling the telemetry from the core functions ## Summary - added callback `experimental_onStart` that exposes the data/events that happen at the very beginning - added callback `experimental_onStepStart` - added callback `experimental_onToolCallStart` - added callback `experimental_onToolCallFinish` - modified callback `onStepFinish` - turned into an event that returns [this](https://github.com/vercel/ai/pull/12654/changes#diff-d56ad94c7c3802f6ed5388ec6735018d179fa70735a40c8c79def79e5927d946R1024-R1045) information ## Manual Verification ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work - add callbacks for streamText() - refactor generateImage - refactor embed - https://github.com/vercel/ai/pull/12654/changes#r2829080416

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## ai@6.0.93 ### Patch Changes - d3769ec: feat(ai): add experimental callbacks in generateText ## @ai-sdk/angular@2.0.94 ### Patch Changes - Updated dependencies [d3769ec] - ai@6.0.93 ## @ai-sdk/langchain@2.0.99 ### Patch Changes - Updated dependencies [d3769ec] - ai@6.0.93 ## @ai-sdk/llamaindex@2.0.93 ### Patch Changes - Updated dependencies [d3769ec] - ai@6.0.93 ## @ai-sdk/react@3.0.95 ### Patch Changes - Updated dependencies [d3769ec] - ai@6.0.93 ## @ai-sdk/rsc@2.0.93 ### Patch Changes - Updated dependencies [d3769ec] - ai@6.0.93 ## @ai-sdk/svelte@4.0.93 ### Patch Changes - Updated dependencies [d3769ec] - ai@6.0.93 ## @ai-sdk/vue@3.0.93 ### Patch Changes - Updated dependencies [d3769ec] - ai@6.0.93 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

…with Output (#12707) ## background preparing for v7 pre-releases by removing all usage of deprecated `generateObject` and `streamObject` from examples, docs, and cookbooks. the core implementation in `packages/ai/src/generate-object/` stays as-is for backwards compat ## summary - delete all `generate-object/` and `stream-object/` example files (equivalents already exist in `generate-text/` and `stream-text/`) - migrate e2e tests, framework examples (next, angular, nuxt, sveltekit) to `generateText`/`streamText` with `Output` - delete `generateObject` and `streamObject` reference pages - remove legacy section from structured data docs - update ~40 doc files (workflows, testing, tool calling, telemetry, troubleshooting, RSC, etc.) - update 19 cookbook files - update 22 provider doc files - add deprecation notes to telemetry docs for old span names api mapping: | before | after | |--------|-------| | `generateObject({ schema })` | `generateText({ output: Output.object({ schema }) })` | | `streamObject({ schema })` | `streamText({ output: Output.object({ schema }) })` | | `output: 'array'` + `schema` | `output: Output.array({ element: schema })` | | `output: 'enum'` + `enum` | `output: Output.choice({ options })` | | `output: 'no-schema'` | `output: Output.json()` | | `result.object` | `result.output` | | `partialObjectStream` | `partialOutputStream` | ## checklist - [x] tests have been added / updated (for bug fixes / features) - [x] documentation has been added / updated (for bug fixes / features) - [ ] a _patch_ changeset for relevant packages has been added (run `pnpm changeset` in root) - [x] i have reviewed this pull request (self-review) ## related issues fixes #12380

## summary - update ai/core area label to reference `Output` and note `generateObject`/`streamObject` as deprecated - stricter `reproduction provided` criteria - requires actual code snippets, not just prose descriptions - prevent tigent from applying `pull request welcome` and `good first issue` labels (humans only) - prevent applying both `bug` and `support` to the same issue ## checklist - [ ] tests have been added / updated (for bug fixes / features) - [ ] documentation has been added / updated (for bug fixes / features) - [ ] a _patch_ changeset for relevant packages has been added (run `pnpm changeset` in root) - [x] i have reviewed this pull request (self-review)

## Background  enabling video models ## Summary  added video models to settings file ## Manual Verification  n/a ## Checklist  - [ ] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work  ## Related Issues

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## ai@6.0.94 ### Patch Changes - Updated dependencies [363fa44] - @ai-sdk/gateway@3.0.52 ## @ai-sdk/angular@2.0.95 ### Patch Changes - ai@6.0.94 ## @ai-sdk/gateway@3.0.52 ### Patch Changes - 363fa44: chore(provider/gateway): update gateway model settings files ## @ai-sdk/langchain@2.0.100 ### Patch Changes - ai@6.0.94 ## @ai-sdk/llamaindex@2.0.94 ### Patch Changes - ai@6.0.94 ## @ai-sdk/react@3.0.96 ### Patch Changes - ai@6.0.94 ## @ai-sdk/rsc@2.0.94 ### Patch Changes - ai@6.0.94 ## @ai-sdk/svelte@4.0.94 ### Patch Changes - ai@6.0.94 ## @ai-sdk/vue@3.0.94 ### Patch Changes - ai@6.0.94 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

…ma (#12471) ## Background PR #12384 introduced support for Anthropic's compaction feature, but the streaming Zod schema for compaction_delta events requires content to be a non-nullable string. Anthropic's API sends compaction_delta events with content: null (e.g. the initial frame before the compaction summary text), which causes Zod validation to fail at runtime. ## Summary Changed content: z.string() to content: z.string().nullish() in the compaction_delta streaming schema in packages/anthropic/src/anthropic-messages-api.ts. This matches the existing content_block_start schema for compaction, which already uses .nullish(). ## Manual Verification Reproduction requires streaming a conversation that exceeds the compaction trigger threshold (50k+ tokens) with @ai-sdk/anthropic, which triggers a compaction_delta event with content: null. This is not practical to verify without an API key and a large conversation, but the added regression test simulates this exact scenario by streaming a compaction_delta chunk with content: null. ## Checklist - Tests have been added / updated (for bug fixes / features) - Documentation has been added / updated (for bug fixes / features) - A patch changeset for relevant packages has been added (for bug fixes / features - run pnpm changeset in the project root) - I have reviewed this pull request (self-review) ## Related Issues Fixes #12470 --------- Co-authored-by: Embedder <215220128+embedder-dev@users.noreply.github.com> Co-authored-by: Aayush Kapoor <83492835+aayush-kapoor@users.noreply.github.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## @ai-sdk/amazon-bedrock@4.0.62 ### Patch Changes - Updated dependencies [b094c07] - @ai-sdk/anthropic@3.0.46 ## @ai-sdk/anthropic@3.0.46 ### Patch Changes - b094c07: fix compaction_delta streaming schema to allow null content ## @ai-sdk/google-vertex@4.0.61 ### Patch Changes - Updated dependencies [b094c07] - @ai-sdk/anthropic@3.0.46 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

…structure routes based on functionality (#12715) ## Summary The `examples/next-openai` example needed clearer naming and better organization to reflect that it's a comprehensive end-to-end example across multiple AI providers. Additionally, most pages and their routes were chat based, so they were moved to `app/chat/` and `app/api/chat/`, with the relevant `chat-` or `use-chat-` prefixes removed from the route names instead. We can do further restructuring as needed, but that seemed enough for now. - **Renamed directory**: `examples/next-openai` → `examples/ai-e2e-next` for clarity - **Agent organization**: Agent implementations reorganized under `agent/<provider>/` folders (e.g., `agent/anthropic/`, `agent/openai/`) - **API route consolidation**: All chat-related API routes moved from `/api/<name>` to `/api/chat/<name>` for proper scoping - **Updated imports**: All relative paths and imports updated to reflect new structure This improves code organization and makes it easier to navigate the example project. ## Manual Verification No manual verification needed for a restructure-only change. All file moves are tracked by git and import paths are updated accordingly. ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [N/A] A _patch_ changeset for relevant packages has been added (examples-only change) - [x] I have reviewed this pull request (self-review) ## Related Issues #12269, #12270 --------- Co-authored-by: Vercel <vercel[bot]@users.noreply.github.com>

Fixes broken link

…ntext editing (#12547) ## Background Anthropic context management support (`clear_tool_uses`, `clear_thinking`) was added in #10540 (with a related fix in #12154), and compaction support (`compact`) was added in #12384. ## Summary With the complexities of handling these features correctly (including how `clear_tool_uses` and `clear_thinking` can be combined), I think it makes sense to add some more tests and also examples around this. - Adds fixture based tests for context editing (tool uses, thinking, and both combined), each for `generateText` and `streamText` - Adds examples for thinking, compaction, and thinking and tool uses combined (there already was an example for tool uses alone, which I renamed for consistency) - Expands docs, adding "Context Editing" grouping section for both `clear_thinking` and `clear_tool_uses` ## Manual Verification All examples were run against the live Anthropic API to verify correct behavior, and the fixture data was captured from those real API responses. ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [ ] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work N/A ## Related Issues N/A

…artbeat keep-alive (#12729) ## Background Video generation requests can take a long time to complete, and HTTP connections may timeout during processing. Server-Sent Events (SSE) with heartbeat keep-alive messages provide a way to maintain the connection while the video is being generated. ## Changes Added SSE support for video generation in the gateway provider with heartbeat keep-alive functionality. The implementation: - Sets `accept: 'text/event-stream'` header in video generation requests - Parses SSE events using `parseJsonEventStream` with a discriminated union schema for result and error events - Properly handles SSE error events by throwing `APICallError` with structured error data ## Manual Verification Ran example video scripts against a gateway server supporting SSE format response. ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work ## Related Issues

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## ai@6.0.95 ### Patch Changes - Updated dependencies [73b7e09] - @ai-sdk/gateway@3.0.53 ## @ai-sdk/angular@2.0.96 ### Patch Changes - ai@6.0.95 ## @ai-sdk/gateway@3.0.53 ### Patch Changes - 73b7e09: feat (provider/gateway): add SSE support for video generation with heartbeat keep-alive ## @ai-sdk/langchain@2.0.101 ### Patch Changes - ai@6.0.95 ## @ai-sdk/llamaindex@2.0.95 ### Patch Changes - ai@6.0.95 ## @ai-sdk/react@3.0.97 ### Patch Changes - ai@6.0.95 ## @ai-sdk/rsc@2.0.95 ### Patch Changes - ai@6.0.95 ## @ai-sdk/svelte@4.0.95 ### Patch Changes - ai@6.0.95 ## @ai-sdk/vue@3.0.95 ### Patch Changes - ai@6.0.95 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

@lgrammel

…to `BedrockChatModelId` (#12657) For some reason, the latest Anthropic models (starting with some from their 4.5 models) were missing from `BedrockChatModelId`. They _were_ however present in `BedrockAnthropicModelId`. Not sure why. When I added support for Opus 4.6 and Sonnet 4.6, I realized this but left it untouched because I wasn't sure whether the previous omissions (of e.g. Opus 4.5) were intentional. Opening this PR to fix it, or please let me know if this was _not_ just an oversight from before. :) cc @lgrammel @gr2m

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## @ai-sdk/amazon-bedrock@4.0.63 ### Patch Changes - dc34ced: fix(provider/amazon-bedrock): add missing latest Anthropic model IDs to `BedrockChatModelId` Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background as a follow up to #12654, we will introduce the same callbacks for streamText function ## Summary - added callback `experimental_onStart` that exposes the data/events that happen at the very beginning - added callback `experimental_onStepStart` - added callback `experimental_onToolCallStart` - added callback `experimental_onToolCallFinish` ## Manual Verification ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work - add similar callbacks for the agent loop

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. # Releases ## ai@6.0.96 ### Patch Changes - 30c9de6: feat(ai): experimental callbacks for streamText ## @ai-sdk/angular@2.0.97 ### Patch Changes - Updated dependencies [30c9de6] - ai@6.0.96 ## @ai-sdk/langchain@2.0.102 ### Patch Changes - Updated dependencies [30c9de6] - ai@6.0.96 ## @ai-sdk/llamaindex@2.0.96 ### Patch Changes - Updated dependencies [30c9de6] - ai@6.0.96 ## @ai-sdk/react@3.0.98 ### Patch Changes - Updated dependencies [30c9de6] - ai@6.0.96 ## @ai-sdk/rsc@2.0.96 ### Patch Changes - Updated dependencies [30c9de6] - ai@6.0.96 ## @ai-sdk/svelte@4.0.96 ### Patch Changes - Updated dependencies [30c9de6] - ai@6.0.96 ## @ai-sdk/vue@3.0.96 ### Patch Changes - Updated dependencies [30c9de6] - ai@6.0.96 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

…s docs (#12741) ## Background Confident AI is an LLM evaluation and observability framework who recently shipped their latest AI SDK observability integration through opentelemetry. ## Summary Updated `content/providers/05-observability/` for: - Adding Confident AI as a provider in the observability integration docs - Followed alphabetical order currently being maintained in the index file ## Checklist - [ ] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [ ] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) --------- Co-authored-by: Aayush Kapoor <83492835+aayush-kapoor@users.noreply.github.com>

## Background as a follow up to #12654, we will introduce the same callbacks for the `ToolLoopAgent` ## Summary - added callback `experimental_onStart` that exposes the data/events that happen at the very beginning - added callback `experimental_onStepStart` - added callback `experimental_onToolCallStart` - added callback `experimental_onToolCallFinish` ## Manual Verification ## Future Work - callbacks aren't currently passed through for streaming - will need to merge #12708 first ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review)

## background the openai provider already defaults to the responses api when calling `openai('model-id')`. the xai provider still defaults to the chat api, which is inconsistent. since this is v7, we can make this breaking change xAI has marked chat completions as a legacy endpoint: https://docs.x.ai/developers/model-capabilities/legacy/chat-completions ## summary - change `xai('model-id')` and `xai.languageModel('model-id')` to use the responses api by default - `xai.chat('model-id')` still available for the chat completions api - update provider interface types to use `XaiResponsesModelId` for default/languageModel - add `chat.ts` examples for generate-text and stream-text - update docs to reflect the new default and link to legacy chat completions api ## verification all xai examples tested against the live api - basic, tool-call, usage, structured-output, search, logprobs, streaming all work with the new default ## checklist - [x] tests have been added / updated (for bug fixes / features) - [x] documentation has been added / updated (for bug fixes / features) - [x] a _patch_ changeset for relevant packages has been added (run `pnpm changeset` in root) - [x] i have reviewed this pull request (self-review) ## related issues closes #13280

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## @ai-sdk/xai@4.0.0-beta.6 ### Major Changes - f62681f: feat(provider/xai): make responses api the default Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## ai@7.0.0-beta.15 ### Patch Changes - Updated dependencies [980f777] - Updated dependencies [7185ba2] - @ai-sdk/gateway@4.0.0-beta.9 ## @ai-sdk/angular@3.0.0-beta.15 ### Patch Changes - ai@7.0.0-beta.15 ## @ai-sdk/gateway@4.0.0-beta.9 ### Patch Changes - 980f777: chore(provider/gateway): update gateway model settings files - 7185ba2: bump `@vercel/oidc` dependency to 3.2.0 ## @ai-sdk/langchain@3.0.0-beta.15 ### Patch Changes - ai@7.0.0-beta.15 ## @ai-sdk/llamaindex@3.0.0-beta.15 ### Patch Changes - ai@7.0.0-beta.15 ## @ai-sdk/react@4.0.0-beta.15 ### Patch Changes - ai@7.0.0-beta.15 ## @ai-sdk/rsc@3.0.0-beta.16 ### Patch Changes - ai@7.0.0-beta.15 ## @ai-sdk/svelte@5.0.0-beta.15 ### Patch Changes - ai@7.0.0-beta.15 ## @ai-sdk/vue@4.0.0-beta.15 ### Patch Changes - ai@7.0.0-beta.15 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background The oauth impl in `@ai-sdk/mcp` had a missing protection check / bug. the `state` parameter which exists to prevent cross-site request forgery during OAuth flows -- was generated and sent to the authorization server, but never validated when the callback came back. this meant an application would accept the callback without noticing the state mismatch and potentially link the victim's tokens to an attacker's session. in practice, `PKCE` (which we implement) would block the specific attack scenario demonstrated in the POC, as long as the OAuth server enforces PKCE validation. But state and PKCE protect against different threat models, and the OAuth spec recommends both. ## Summary - added `saveState()` and `storedState()` to OAuthClientProvider - added `callbackState` to `auth()` - applications can now pass the state they received from the OAuth callback URL - state is saved when starting a flow - state is validated before token exchange ## Manual Verification na ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review)

## Background forgot to add changeset for #13384 ## Summary added changeset ## Manual Verification na ## Checklist - [ ] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review)

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## @ai-sdk/mcp@2.0.0-beta.3 ### Patch Changes - b9b3899: changeset for #13384 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background maintenance work for consistency across changelogs

## Background Currently, OpenTelemetry is tightly integrated into the the AI functions as our only mechanism to deliver telemetry data. With the new telemetry system, we want to separate specific implementations such as OTel from the AI functions themselves. ## Summary Extract the OTel span attribute creation and updating into a dedicated OTel telemetry integration that uses our callbacks and is automatically part of the global telemetry. All OTel specific code is removed from `generateText`, `streamText`, and related files. Additional callbacks `onError` and `onChunk` are added to support the required behavior. ## Breaking Changes * telemetry settings tracer is no longer used in `streamText` and `generateText` ## Manual Verification - ran the example `examples/ai-functions/src/generate-text/anthropic-with-langfuse-telemetry.ts` ## Future Work - move otel into separate package and remove from global registration (users need to manually set it up) - add similar telemetry architecture to other ai functions - after that, remove tracer from telemetrysettings - new telemetry documentation ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [ ] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) --------- Co-authored-by: Lars Grammel <lars.grammel@gmail.com>

This is an automated update of the gateway model settings files. Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## ai@7.0.0-beta.16 ### Major Changes - 5b8c58f: feat(ai): decouple otel from core functions ## @ai-sdk/angular@3.0.0-beta.16 ### Patch Changes - Updated dependencies [5b8c58f] - ai@7.0.0-beta.16 ## @ai-sdk/langchain@3.0.0-beta.16 ### Patch Changes - Updated dependencies [5b8c58f] - ai@7.0.0-beta.16 ## @ai-sdk/llamaindex@3.0.0-beta.16 ### Patch Changes - Updated dependencies [5b8c58f] - ai@7.0.0-beta.16 ## @ai-sdk/react@4.0.0-beta.16 ### Patch Changes - Updated dependencies [5b8c58f] - ai@7.0.0-beta.16 ## @ai-sdk/rsc@3.0.0-beta.17 ### Patch Changes - Updated dependencies [5b8c58f] - ai@7.0.0-beta.16 ## @ai-sdk/svelte@5.0.0-beta.16 ### Patch Changes - Updated dependencies [5b8c58f] - ai@7.0.0-beta.16 ## @ai-sdk/vue@4.0.0-beta.16 ### Patch Changes - Updated dependencies [5b8c58f] - ai@7.0.0-beta.16 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## ai@7.0.0-beta.17 ### Patch Changes - Updated dependencies [4d6ab9a] - @ai-sdk/gateway@4.0.0-beta.10 ## @ai-sdk/angular@3.0.0-beta.17 ### Patch Changes - ai@7.0.0-beta.17 ## @ai-sdk/gateway@4.0.0-beta.10 ### Patch Changes - 4d6ab9a: chore(provider/gateway): update gateway model settings files ## @ai-sdk/langchain@3.0.0-beta.17 ### Patch Changes - ai@7.0.0-beta.17 ## @ai-sdk/llamaindex@3.0.0-beta.17 ### Patch Changes - ai@7.0.0-beta.17 ## @ai-sdk/react@4.0.0-beta.17 ### Patch Changes - ai@7.0.0-beta.17 ## @ai-sdk/rsc@3.0.0-beta.18 ### Patch Changes - ai@7.0.0-beta.17 ## @ai-sdk/svelte@5.0.0-beta.17 ### Patch Changes - ai@7.0.0-beta.17 ## @ai-sdk/vue@4.0.0-beta.17 ### Patch Changes - ai@7.0.0-beta.17 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background the package providers still relied on V3 specs instead of the v4 ones (for main branch) ## Summary - update to use v4 spec ## Manual Verification na ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review)

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## @ai-sdk/amazon-bedrock@5.0.0-beta.3 ### Patch Changes - e748b35: chore: update v3 specs to v4 - Updated dependencies [e748b35] - @ai-sdk/anthropic@4.0.0-beta.3 ## @ai-sdk/anthropic@4.0.0-beta.3 ### Patch Changes - e748b35: chore: update v3 specs to v4 ## @ai-sdk/google-vertex@5.0.0-beta.10 ### Patch Changes - e748b35: chore: update v3 specs to v4 - Updated dependencies [e748b35] - @ai-sdk/anthropic@4.0.0-beta.3 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background the package providers still relied on V3 specs instead of the v4 ones (for main branch) ## Summary - update to use v4 spec ## Manual Verification na ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review)

## background xAI responses API supports `reasoning.summary` to control how reasoning traces are presented, but the SDK only exposed `reasoning.effort` ## summary - add `reasoningSummary` to responses provider options - map to `reasoning.summary` in request body alongside existing `reasoning.effort` - add tests for summary alone, and effort + summary together - add generate-text and stream-text examples ## verification <details> <summary>reasoning.summary sent alongside reasoning.effort</summary> ``` pnpm tsx src/generate-text/xai/responses-reasoning-summary.ts Quantum entanglement is a phenomenon in quantum mechanics where two or more particles become interconnected... Usage: { outputTokenDetails: { textTokens: 104, reasoningTokens: 299 }, ... } ``` </details> ## checklist - [x] tests have been added / updated (for bug fixes / features) - [ ] documentation has been added / updated (for bug fixes / features) - [x] a _patch_ changeset for relevant packages has been added (run `pnpm changeset` in root) - [x] i have reviewed this pull request (self-review) ## related issues part of #12825

…oGenerate (#13372) ## background xAI's non-streaming responses API returns reasoning text in the `content` field on reasoning output items (as `reasoning_text` parts), not in the `summary` field. the SDK only read from `summary`, so reasoning was silently dropped in `generateText` - streaming worked fine since it uses `reasoning_summary_text.delta` events ## summary - add `content` field to the reasoning output item schema in `xai-responses-api.ts` - read reasoning text from `content` when `summary` is empty in doGenerate - add test for the content fallback path - add generate-text reasoning example ## verification <details> <summary>before: no reasoning shown</summary> ``` $ pnpm tsx src/generate-text/xai/responses-reasoning.ts --- text --- The word "strawberry" contains **3** instances of the letter "r". ``` </details> <details> <summary>after: reasoning properly extracted</summary> ``` $ pnpm tsx src/generate-text/xai/responses-reasoning.ts --- reasoning --- First, the user asked: "How many 'r's are in the word 'strawberry'?" I need to count the number of times the letter 'r' appears in the word "strawberry". Let me spell out the word: S-T-R-A-W-B-E-R-R-Y. Now, let's go through each letter: 1. S - Not an 'r' 2. T - Not an 'r' 3. R - This is an 'r' (first one) 4. A - Not an 'r' 5. W - Not an 'r' 6. B - Not an 'r' 7. E - Not an 'r' 8. R - This is an 'r' (second one) 9. R - This is an 'r' (third one) 10. Y - Not an 'r' So, the 'r's are at positions 3, 8, and 9. That means there are 3 'r's in "strawberry". --- text --- The word "strawberry" contains **3** instances of the letter "r". ``` </details> ## checklist - [x] tests have been added / updated (for bug fixes / features) - [ ] documentation has been added / updated (for bug fixes / features) - [x] a _patch_ changeset for relevant packages has been added (run `pnpm changeset` in root) - [x] i have reviewed this pull request (self-review) --------- Co-authored-by: Felix Arntz <felix.arntz@vercel.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## @ai-sdk/alibaba@2.0.0-beta.3 ### Patch Changes - 4ab27b9: chore(alibaba): update v3 specs to v4 ## @ai-sdk/xai@4.0.0-beta.7 ### Patch Changes - 0f11f10: add reasoningSummary to responses API provider options - 12115e9: fix reasoning text extraction from content in responses doGenerate Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

## Background the package providers still relied on V3 specs instead of the v4 ones (for main branch) ## Summary - update to use v4 spec ## Manual Verification na ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review)

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## @ai-sdk/google@4.0.0-beta.9 ### Patch Changes - e59176c: chore(google): update v3 specs to v4 ## @ai-sdk/google-vertex@5.0.0-beta.11 ### Patch Changes - e59176c: chore(google): update v3 specs to v4 - Updated dependencies [e59176c] - @ai-sdk/google@4.0.0-beta.9 Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

This PR contains the following updates: | Package | Change | [Age](https://docs.renovatebot.com/merge-confidence/) | [Confidence](https://docs.renovatebot.com/merge-confidence/) | |---|---|---|---| | [pkce-challenge](https://redirect.github.com/crouchcd/pkce-challenge) | [`5.0.0` → `5.0.1`](https://renovatebot.com/diffs/npm/pkce-challenge/5.0.0/5.0.1) | ![age](https://developer.mend.io/api/mc/badges/age/npm/pkce-challenge/5.0.1?slim=true) | ![confidence](https://developer.mend.io/api/mc/badges/confidence/npm/pkce-challenge/5.0.0/5.0.1?slim=true) | --- ### Release Notes <details> <summary>crouchcd/pkce-challenge (pkce-challenge)</summary> ### [`v5.0.1`](https://redirect.github.com/crouchcd/pkce-challenge/blob/HEAD/CHANGELOG.md#501---2025-11-22) [Compare Source](https://redirect.github.com/crouchcd/pkce-challenge/compare/5.0.0...5.0.1) </details> --- ### Configuration 📅 **Schedule**: Branch creation - Between 05:00 AM and 05:59 AM, only on Friday ( * 5 * * 5 ) in timezone America/Los_Angeles, Automerge - At any time (no schedule defined). 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this PR and you won't be reminded about this update again. --- - [ ] If you want to rebase/retry this PR, check this box --- This PR was generated by [Mend Renovate](https://mend.io/renovate/). View the [repository job log](https://developer.mend.io/github/vercel/ai).  --------- Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> Co-authored-by: Gregor Martynus <39992+gr2m@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

## Background this doc is essentially to track all the breaking changes that may happen in v7 of AISDK Remember to make changes to this doc whenever a breaking change is introduced ## Summary - added a migration guide to track the breaking change of passing a `tracer` via the `integrations` option and not via the `tracer` option ## Manual Verification na ## Checklist - [ ] Tests have been added / updated (for bug fixes / features) - [x] Documentation has been added / updated (for bug fixes / features) - [ ] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review)

## Background As part of the ongoing migration from V3 to V4 provider specification types, the `xai` package still referenced V3 types from `provider`. ## Summary Similar to e.g. #13375, this updates the xAI provider to use the V4 spec types. ## Manual Verification N/A ## Checklist - [x] Tests have been added / updated (for bug fixes / features) - [ ] Documentation has been added / updated (for bug fixes / features) - [x] A _patch_ changeset for relevant packages has been added (for bug fixes / features - run `pnpm changeset` in the project root) - [x] I have reviewed this pull request (self-review) ## Future Work N/A ## Related Issues N/A

This PR contains the following updates: | Package | Change | [Age](https://docs.renovatebot.com/merge-confidence/) | [Confidence](https://docs.renovatebot.com/merge-confidence/) | |---|---|---|---| | [@smithy/util-utf8](https://redirect.github.com/smithy-lang/smithy-typescript/tree/main/packages/util-utf8) ([source](https://redirect.github.com/smithy-lang/smithy-typescript/tree/HEAD/packages/util-utf8)) | [`4.0.0` → `4.2.1`](https://renovatebot.com/diffs/npm/@smithy%2futil-utf8/4.0.0/4.2.1) | ![age](https://developer.mend.io/api/mc/badges/age/npm/@smithy%2futil-utf8/4.2.1?slim=true) | ![confidence](https://developer.mend.io/api/mc/badges/confidence/npm/@smithy%2futil-utf8/4.0.0/4.2.1?slim=true) | --- ### Release Notes <details> <summary>smithy-lang/smithy-typescript (@smithy/util-utf8)</summary> ### [`v4.2.1`](https://redirect.github.com/smithy-lang/smithy-typescript/blob/HEAD/packages/util-utf8/CHANGELOG.md#421) [Compare Source](https://redirect.github.com/smithy-lang/smithy-typescript/compare/@smithy/util-utf8@4.2.0...@smithy/util-utf8@4.2.1) ##### Patch Changes - [`03c3dc8`](https://redirect.github.com/smithy-lang/smithy-typescript/commit/03c3dc8): update for rollup build externalLiveBindings=false - Updated dependencies \[[`03c3dc8`](https://redirect.github.com/smithy-lang/smithy-typescript/commit/03c3dc8)] - [@smithy/util-buffer-from](https://redirect.github.com/smithy/util-buffer-from)@4.2.1 ### [`v4.2.0`](https://redirect.github.com/smithy-lang/smithy-typescript/blob/HEAD/packages/util-utf8/CHANGELOG.md#420) [Compare Source](https://redirect.github.com/smithy-lang/smithy-typescript/compare/@smithy/util-utf8@4.1.0...@smithy/util-utf8@4.2.0) ##### Minor Changes - [`45ee67f`](https://redirect.github.com/smithy-lang/smithy-typescript/commit/45ee67f): update dist-cjs generation to use rollup ##### Patch Changes - Updated dependencies \[[`45ee67f`](https://redirect.github.com/smithy-lang/smithy-typescript/commit/45ee67f)] - [@smithy/util-buffer-from](https://redirect.github.com/smithy/util-buffer-from)@4.2.0 ### [`v4.1.0`](https://redirect.github.com/smithy-lang/smithy-typescript/blob/HEAD/packages/util-utf8/CHANGELOG.md#410) [Compare Source](https://redirect.github.com/smithy-lang/smithy-typescript/compare/@smithy/util-utf8@4.0.0...@smithy/util-utf8@4.1.0) ##### Minor Changes - [`64cda93`](https://redirect.github.com/smithy-lang/smithy-typescript/commit/64cda93): set sideEffects bundler metadata ##### Patch Changes - Updated dependencies \[[`64cda93`](https://redirect.github.com/smithy-lang/smithy-typescript/commit/64cda93)] - [@smithy/util-buffer-from](https://redirect.github.com/smithy/util-buffer-from)@4.1.0 </details> --- ### Configuration 📅 **Schedule**: Branch creation - Between 05:00 AM and 05:59 AM, only on Friday ( * 5 * * 5 ) in timezone America/Los_Angeles, Automerge - At any time (no schedule defined). 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this PR and you won't be reminded about this update again. --- - [ ] If you want to rebase/retry this PR, check this box --- This PR was generated by [Mend Renovate](https://mend.io/renovate/). View the [repository job log](https://developer.mend.io/github/vercel/ai).  --------- Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> Co-authored-by: Gregor Martynus <39992+gr2m@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

This PR was opened by the [Changesets release](https://github.com/changesets/action) GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated. ⚠️

⚠️

⚠️ `main` is currently in **pre mode** so this branch has prereleases rather than normal releases. If you want to exit prereleases, run `changeset pre exit` on `main`. ⚠️

⚠️

⚠️ # Releases ## @ai-sdk/xai@4.0.0-beta.8 ### Patch Changes - 4799fa5: chore(provider/xai): update provider to use v4 types Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

aayush-kapoor and others added 30 commits February 18, 2026 11:25

Update tigent.yml to clarify label usage (#12681)

eae7943

see #12673. Agents shouldn't apply these labels, they are reserved for humans to control releases and backports. I really don't want to accidentally release a breaking change lol

fix (docs): broken agents cookbook link (#12734)

efc937c

Fixes broken link

dancer and others added 30 commits March 12, 2026 09:47

chore(provider/gateway): update gateway model settings files (#13360)

980f777

update oidc (#12444)

7185ba2

docs(skills): make ADR skill internal (#13381)

896e846

chore: update changelog description for 'mcp' (#13401)

f3d7135

## Background maintenance work for consistency across changelogs

chore(provider/gateway): update gateway model settings files (#13404)

4d6ab9a

This is an automated update of the gateway model settings files. Co-authored-by: vercel-ai-sdk[bot] <225926702+vercel-ai-sdk[bot]@users.noreply.github.com>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

backport v6#5

backport v6#5
callycodes wants to merge 1773 commits intoAllTheTables:mainfrom
vercel:main

callycodes commented Jan 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

20 participants

Conversation

callycodes commented Jan 6, 2026

Background

Summary

Manual Verification

Checklist

Future Work

Related Issues

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

20 participants